Intelligence Augmentation vs. Artificial Intelligence
To some, it seems "obvious" that significant human intelligence augmentation will come before human-level AI. To others, it's the reverse that's obvious. I don't think either is obvious, but I believe there's a strong likelihood AI will come first.
In the IA camp, one of the arguments goes, [Brain+Computer] will always be more intelligent than [Computer] alone. But this is untrue, as the I/O channels between brain and computer make all the difference, and with today's technology, these channels are quite limited. Even if we had million-electrode brain-computer interfaces, it would be a cybernetics problem to ask which outputs to plug into which inputs, and what changes might need to be made to the central executive to handle the new cognitive architecture without information overload or psychosis. Reprogramming the executive center of the human brain would require advanced neurosurgery and extensive knowledge of the brain, knowledge that could take decades of research and advanced experimental techniques to uncover.
Other cons for IA, in my view:
- Experimentation on the human brain is likely to be made illegal globally
- The design-and-test cycle is on the order of weeks or months
- Lack of human volunteers willing to die for the cause of IA research
- Someone left out the line notes for the brain's code
- Experimenting on the deep brain is difficult because neocortex is in the way
- All that medical hardware is really expensive
- The human brain was not designed to be upgraded
- Gene therapies not likely to give enough improvement for takeoff speed
A remark on that last one... the issue of takeoff speed. It's not enough to create an Einstein with IA. You have to create an Einstein that can go immediately to work on new intelligence augmentation techniques, and actually come up with something of use in a reasonable amount of time, before AI is developed. It seems more likely to me that an intelligence-enhanced human would just go into the business of creating AI. Smarter-than-human intelligence cannot just be a really smart human being - it has to be something qualitatively off the scale. Manipulating the genes associated with genius, as James Miller suggested, would likely produce "only" human geniuses at first. You'd need to go an extra level of theory and genetic engineering to get something genuinely smarter-than-human in a human-like package. I'm not saying it couldn't be done, but that the whole process could drag on for a number of years.
Benefits of IA:
- Evolution has already done a lot of work for us
- Some might think a human seed is more predictable
- Sparks human-centric patriotism in ways AI doesn't
On to the cons of AI:
- Present-day computers might not be fast enough to implement AI
- You have to build everything from scratch yourself
- Everyone is working on narrow AI, but AGI is unpopular
- Requires strong theory of general intelligence, difficulty unknown
- Stigma of excessive past claims
And the benefits of AI:
- Design-and-test cycle can be very rapid
- All aspects of the AI are read/write friendly
- Line notes are included with the code
- Cognitive features can be optimized for self-improvement
- Computational power can be expanded as funds allow
- Virtual worlds are available as a flexible training zone
- Hardware can be used to "overclock" beneficial functions
- Probabilistically realistic, flexible learning can be implemented
- Nascent AIs can share information with each other rapidly
- Much larger regions of the mind configuration space can be tested
- AIs can be copied indefinitely, allowing to commercial spin-offs
- Substantial advances in AI, but not IA, have already been achieved
- The hardware itself is inherently cheaper
- Little to no legal concerns
Comment away. Whether or not IA or AI reaches smarter-than-human intelligence first is pretty important, as the step into this new domain could spark a runaway self-improvement process, something I.J. Good called an "intelligence explosion". This is normally what we think of when we hear the word superintelligence.
June 8th, 2007 - 08:44
I cannot concur. Neural plasticity for tool use seems a strongly “upgrade-friendly” characteristic, as the mechanism is not unique to the motor cortex — as is demonstrated by synaesthesia and sensory deprivation/modification adaptation.
I have sentiments in the other areas, but this one statement in particular alone seems patently falsified.
June 8th, 2007 - 08:53
In terms of intelligence augmentation, what is being done now with the most impact and near term potential. Although I think the focus should be performance/productivity/results augmentation.
Revamped education==>Deliberate practice
Here is a paper on achieving expert performance
http://www.psy.fsu.edu/faculty/ericsson/ericsson.exp.perf.html
This is described here as well
changing learning systems (that also tap into automated systems for helping to tap into more information as well as systems for recording and providing feedback on performance) could accelerate the creation of experts and allow for more people to be effectively turned into experts. Performance and productivity could be greatly improved across a broadrange of activities. [blogging actually provides quality feedback on ideas]
Intelligence (IQ) would not be changed but performance and productivity would be.
Nasa has deep learning view of transfering the knowlegdge of experts
http://appel.nasa.gov/ask/issues/22/22_preserving_leonard.php
One of the biggest areas to apply tighter performance monitoring, feedback and expert tutoring would be for government and business leaders. There is some feedback now, especially for business (stock performance and profitability). However, leaders of large corporations and nations can establish entrenched positions that are disconnected from performance consequences. Also, the feedback and system can be reinforcing expertise in non-productive performance (ie how to get there, stay there but not about making good things happen overall.)
the expert creation/performance level cycle can also be shortcutted with artificial aids (wikipedia, google, internet, IM a friend, video conference collaboration, etc..)
I think more performance enhancement is possible with nootropic drugs, gene therapy and also with not necessarily invasive computer interfaces.
http://advancednano.blogspot.com/2006/12/state-of-cognitive-enhancement.html
June 8th, 2007 - 09:36
I’d have to say that despite your excellent summation, I’m in the Intelligence Augmentation camp.
The reason for this is an item that needs inclusion to the con side of AI:
A working understanding of the mechanisms of intelligence are required (creativity, intuitiveness, consciousness, emotion?, etc)from both a substrate and software level.
Once we understand the mechanics of our own intelligence making AI will be more achievable, but so will IA. IA maintains, for the moment, the advantage of a working predicate system.
June 8th, 2007 - 10:44
Substantial advances in AI, but not IA, have already been achieved
Augmenting human intelligence by hardware has been around ever since cave painting was invented. The real question is whether the bandwidth can be increased.
June 8th, 2007 - 12:02
It’s an excruciating dilemma – both to guess which will show results first, and which is the better idea to advocate.
The latter depends on your take on the ethical character of humanity. is the glass half full, or half empty? Are we more good than selfish? Are we more incompetent than cautious?
It’s instinctively easier to trust someone you know (h. sapiens) than someone you not only don’t, but CAN’T know (synthetic AGI), even if it can be made Friendly – and far Friendlier (more compassionate, empathetic, and wise) than any human is capable of.
Usually, the way you solve a prisoner’s dilemma is to get to know the other prisoners. So until people can actually start getting experience interacting with AGIs (Friendly or otherwise), advocates of AGI over IA have something of an uphill battle.
Personally, I’m cynical enough about the human species that I actually don’t believe unequivocally that IA would be better than AGI (we already have humans in charge, and we’re not in all that hot of a shape). Intelligence is not the problem – it’s the motivational structure. You can’t ethically reprogram humans, but you can ethically reprogram AGI.
What I’m concerned about is how much IA we’re going to have to actually develop in order to properly implement Friendliness. Is an FAI going to have to understand us better than we do? Or would it be sufficient to just be motivationally among the more generous of human beings, with the power of recursive self-improvement?
June 8th, 2007 - 12:21
IanC wrote:
The brain does have a large degree of plasticity and adaptiveness, sure. Still, the sentence “the brain was not designed to be upgraded” (or, rather, “the brain did not evolve to be upgraded”) cannot be falsified simply by citing some examples of how we can learn new things. After well, the brain didn’t evolve to be easy to upgrade. It evolved to maximize fitness, like all evolutionary adaptations do. Nothing in maximizing fitness requires you to develop plug’n'play functionality for modules that don’t exist yet.
Regarding the neural plasticity – how about musician’s cramp? (unfortunately couldn’t find a non-gated version) It’s a condition where professional musicians start to mix up sensations coming from their fingers – thinking they touched something with their index finger, for instance, when they actually touched it with their middle finger. While it can be healed, in the past it was a career-ender. To quote the article I linked to:
Using brain images called magnetic encephalograms, his team found that the finger maps in the sensory cortex of the musicians’ brains were smeared together; the fingers sent signals to overlapping cortex areas. While their fingers were thus less coordinated, the musicians kept forcing their way through complex movements–which ultimately left them in pain, Elbert proposes.
Why does intensive practice on an instrument cause the brain’s sensory areas to become scrambled? “What fires together, wires together,” says neuroscientist Michael Merzenich of the University of California, San Francisco. The musicians were probably moving their fingers so quickly that the brain couldn’t tell one finger’s signal from the next, he says.
In other words, neural plasticity causes problems even where it helps things. Phantom limb pain is another example of a case where the vaunted plasticity is more harmful than useful: after a certain part of the brain stops receiving input from the body, its neurons get reorganized to assist in the functions of the neighboring sense centers. It’s too bad that activity in the subsumed center is still perceived as activity in the lost limb – touch somebody on the face, for instance, and they might feel pain in the leg.
Were the brain really “designed to be upgradable”, the neighboring centers would take over the freed up resources cleanly and effectively. In a computer system, shutting down an unnecessary process clears the processor cycles for new use without the computer experiencing “phantom bugs” because the processor was used to doing something else. As long as the system is built modularly, modules and subprograms can be freely removed or added, without worrying about any such concerns. If just started keying in new forms of input into our existing sensory centers, those centers might expand too much, overlapping with neighboring ones with bad consequences. People who lose one sense will have their remaining ones sharpen, suggesting that there’s a very real resource limit on the brain. Give it too many direct computer interfaces, and you’ll start running out of neurons to handle them. Dogs have a keener sense of smell than we do, but dogs have much larger portions of their brain dedicated to processing smells. Depression and other serious brain dysfunctions are caused by comparatively small changes in neurotransmitter levels – once you start doing your immense upgrades, you have to do a very delicate job of making sure they stay at the right levels.
I could go on for several paragraphs, but maybe I’ve made my point. We’re not evolved to be upgradable. (If we were so oriented towards upgrades by nature, how come even the most recent upgrades that nature created aren’t perfectly implemented? I have only the barest conscious control over my emotional system, even my emotions are clearly harming me. Why isn’t that better keyed to my higher-level control, if we’re so clearly designed to be upgraded?)
June 8th, 2007 - 12:40
The brain was not designed at all, of course. Yet the brain can certainly be adapted to work with designed augments. As Professor Wang stated, performance augmentation is well within the reach of most humans, even without measurable increase in IQ.
Too many of the persons speculating about mental augmentation of humans do not have the required background to understand the intricacies involved.
The failure of traditional AI to understand and recreate human level cognition is not merely a lack of sufficiently complex computer hardware, or sufficiently sophisticated algorithms. The failure of understanding is not quantitative, but rather is qualitative. There is not helping it, either.
Are we evolved to be upgradable? Of course not, that is not the point of evolution. However, we are evolved enough to sidestep evolution and conceive of work-around upgrades that could work. Most humans do not understand what intelligence actually is.
Which is why machine intelligence is likely to lag behind augmented human intelligence. Humans are capable of augmenting something that already exists. Until humans understand the concept of intelligence better, however, they will not be able to create a non-human intelligence to challenge augmented humans.
June 8th, 2007 - 12:58
Brian Wang is not a professor… you may be thinking of Pei Wang. Brian certainly is a clever guy, though. ;-)
Not enough people have even tried for AGI in the recent past for your statement on a qualitative failure of understanding to make sense. If you watch the video I posted a few days ago, you’ll see that Ben Goertzel has had some nice successes applying his AI design to analysis in bioinformatics. There are hundreds of other promising examples of AI progress in the last decade or so, and our $1,000 PCs are just beginning to reach 1/100th the processing power of the human brain, so it’s a bit premature to say hardware is not a problem.
IA advocates always leave out the part that true, serious, hands-on experimentation on the human brain is illegal and in many cases unethical… yes, there are better organizational practices and whatnot that can help us get a little bit smarter, but I’m talking about sci-fi-esque human brain upgrades that create supergeniuses. Implementing these in the next two decades is not out of the question, but there are little specific plans to do it.
I’m extremely skeptical of those who are hyperbolically defeatist about either AI, as you are in your comment, or IA, as a few people I know are. There is not “no helping” either of them. Rudimentary AI has made billions of dollars in the last few years. The “AI effect” is that anything AI-like which is useful begins to be called “software”. There are multiple obvious promising routes to IA. But most of them involve deep brain electrodes and boosting the electrode count.
I’ve been looking at both IA and AI in detail for the last seven years, and believe me, neither of them should be written off.
June 8th, 2007 - 13:39
I think I agree with Joseph actually. What if Artificial Intelligence WAS Intelligence Augmentation ? What if there was no need to antagonize these two concepts at all ? What if artefacts always were the continuity of our flesh, bones and souls?
June 8th, 2007 - 14:40
Sure, what’s currently called AI is actually IA. Like Joseph and Metz pointed out, we’re currently using technology to enhance our own cognition, not as an independent agent with a cognition of its own. But we’re talking about superintelligence level, now – the level of technology where including a human in an AI process would be the worst bottleneck ever. Imagine marrying an otherwise top-of-the-line computer setup with a 1 Mhz CPU. At that point, you either turn things completely over to AI systems, or you upgrade yourself to become as fast as they are yourself. If you don’t, you’ll be overrun by the competition who does.
June 8th, 2007 - 15:30
You make a good point Michael.
Aiming for IA is like taking the Windows source code and using that to build the best OS ever.
It’s tempting, because you start out with millions of lines of code.
However, in the end we all know it’s best to just start from scratch.
June 8th, 2007 - 15:32
We may yet see, over the next 8-12 yrs, the convergence/coalescence of IA with A(G)I. But, as I’ve remarked before, and for reasons that Michael & Mitch Howe articulate well, this does NOT at all mitigate against forging ahead with AGI in and of itself—on the contrary, it argues in favor of a strong push in AGI itself (in my judgment, at least).
June 8th, 2007 - 16:29
Because you haven’t practiced properly. Using even the best tools in ways not in convergence with the User’s Manual is a good way to get yourself into trouble. It can be done, by the way. Much in a similar manner to the capacity of some Buddhist monks to survive in analogous-to-sub-arctic conditions in nothing more than a thin linen blanket overnight. :)
Not if we augment the available capacity of the given centers alongside the extra sensory input, much as Dr. Berger is doing with his hippocampal implants as a theoretical remedy for Alzheimer’s.
And that’s the point you missed; the human brain is “designed by evolution” to be highly adaptable. Adaptability is, effectively, upgrade-ability.
June 8th, 2007 - 16:29
Very interesting analogy Jan-Willem! I don’t see it as entirely correct though because I do think that IA has potential… but I question the motivations of many advocates of *only* IA. I think they might have an anti-AI complex. Pro-AI folk, on the other hand, rarely if ever have a specifically pro-AI complex. Many pro-AI folk in fact see AI as the fastest route to substantial IA, which is part of why they want it. This is the view of the SIAI people. The other reason is reliably protecting against UFAI.
Let me be clear that faith in mankind or love of humanity should not factor into our assessments here. We can love humanity and still think that AI will hit true superintelligence before IA. It’s a technical question, not a moral one.
Also note the probability of friendly AI superintelligence vs. the probability of friendly IA superintelligence are also irrelevant to the answer… what we would prefer is not necessarily what’s technologically easiest.
These issues can obfuscate the question…
Ian, no ad hominem please. I would prefer if people have discussions as if they’re both submitting papers to a journal rather than bantering back and forth.
June 8th, 2007 - 17:38
Fact check: “intelligence explosion” was coined by I. J. Good, right? Eli and Google seem to agree.
Also, “virtual worlds” isn’t necessarily distinctly an AI advantage.
Overall a good article, preaching to the choir. :)
Would you elaborate on this?
June 8th, 2007 - 17:57
Thanks, I meant I.J. Good.
The more electrodes you have the bigger your I/O bandwidth is. This is fundamental to anyone working on neuroprostheses or brain-computer interfacing.
June 8th, 2007 - 18:34
Michael — Duly noted. I wouldn’t call what I did there ad hominem, though; in the former instance I was addressing a personal analogy and in the second, I wasn’t assaulting his character in any way shape or form. But even so, I’ll try to “behave” better. :)
June 8th, 2007 - 19:06
“Because you haven’t practiced properly.”
Really? And so we can achieve complete control over our emotions and autonomous bodily functions if we just practice enough? I wonder why no army in the world has thought of this; you could use this to train super-soldiers who would keep fighting just as efficiently even when in severe pain.
“Using even the best tools in ways not in convergence with the User’s Manual is a good way to get yourself into trouble.”
I didn’t know evolution designed a User’s Manual for humans. Are the manual’s procedures included in ISO 9001?
“It can be done, by the way.”
Weren’t you going on and on about how I didn’t present enough evidence for my claims just a few articles ago?
“Much in a similar manner to the capacity of some Buddhist monks to survive in analogous-to-sub-arctic conditions in nothing more than a thin linen blanket overnight.”
Emotions have very little to do with that; it’s simply the body’s predesigned response mechanisms to cold. It’s not like we can take the body and redesign it so it can operate for hours on end in -40 C weather without heavy protection from the elements.
“Adaptability is, effectively, upgrade-ability.”
My PC is very highly adaptable- it can run over ten thousand different predesigned software programs and any code which I can write. No sane person would think of these as being equivalent to hardware upgrades. You’re kind of missing the point- an upgraded human, by definition, can do things no regular human ever could even with years of training. The idea behind upgraded humans is that we’ll be able to do entirely new stuff that nobody has ever done before.
June 8th, 2007 - 20:05
Ian,
Saying “because YOU haven’t practiced properly” with regards to lack of emotional control – a humanity-wide issue – is ad hominem. It is fantasically obvious that human control over our emotional systems is limited, and even the best monk in the world only has a limited degree of control over the activation of their own emotions. You acting like Kaj has a special problem, that he can’t control his own emotions because he hasn’t tried, which is totally ad hominem. Tom McCabe’s point about super-soldiers also underscores the point. Humans do not have complete control over our own emotions, period. If it were possible to do this then such individuals would be blatantly superior decision-makers and would be paid thousands of bucks an hour, most likely. There would be numerous scientific papers on the cognitive differences between these magical emotion-controlling people and the non-emotion-controllers.
Anyway, read what Tom McCabe has to say and don’t be afraid to slightly retune your beliefs – ever so slightly – if you think the arguments are persuasive that is!
And btw Tom…
Shouldn’t you be linking your new blog in your comments? ;-)
June 8th, 2007 - 20:51
No matter which comes first: there will be synergies where both fields feed off each other.
I for my part am convinced that increases in the ability to watch brain processes in real time and thus, through neurophysiological insights, advances in the understanding of human intelligence will be crucial to any attempt at AI.
As much as I believe the work by Yudkowsy and others on AGI is important I still don’t believe that any significant breakthrough will be made without decisive help from “reverse engineering” the human brain.
June 8th, 2007 - 21:44
Return on investment. It takes decades of training in severely austere conditions. It also precludes the maintenance of military skills & reflexes.
Now was ad hominem — Tom, it should be painfully obvious that I wasn’t being literal.
Michael — in all honesty, I again cannot agree with you on this subject. It is possible to control one’s emotional responses, just as it is possible to control one’s autonomous responses. It takes the same level of dedication and effort, however; a level which is beyond the realm of normal, ordinary people — only the “truly dedicated” to that task can accomplish it. I know that it can be done, however, as I have experience in both areas. The autonomous due to a lifetime of being in pain from severe migraines; the emotional due to other concerns that are somewhat related to that, and a history of severe anger issues. Again: It can be done; with that in mind, my comment was not ad hominem in the attack sense. Especially since, while personally directed, it was in response to a personally framed question.
No, Tom: I was saying you needed to provide evidence to make claims with. There’s a difference there. Nothing I ever saw was that.
June 8th, 2007 - 21:45
What a coincidence:
I just found this article at PhysOrg.com – it neatly illustrates what I meant in my comment above:
“Research deciphers ‘déjà -vu’ brain mechanics” (http://www.physorg.com/news100535450.html)
June 8th, 2007 - 22:15
On Emotion Control.
On Physiological Control:
The “Sub-arctic” bit was exaggeration, yeah — ya got me there. But the point is illustrated successfully nonetheless.
June 8th, 2007 - 22:19
Since I’m apparently spamming:
If you’re going to call me on it, Michael, you need to call Tom as well.
And to Tom’s point here: there is no such thing as a software/hardware distinction to the human brain. Adaptability in one is adaptability in the other. Adaptable hardware is upgradeable hardware. Proof-of-concept via logic.
June 8th, 2007 - 22:56
Despite the fact that AI is, as of now, “the unknown,” it seems that it would be, on the advent of its actual creation, vastly more know-able, being a highly supervised and detailed creation made by a human – or more likely, group of humans – rather than something that simply emerged as a result of random chance. If someone were to create a low-level (non-super?) AGI, wouldn’t it be likely that the creator know more about that intelligence than he would about human intelligence? (I specify non-super because I have to assume that, without IA, humans would eventually fall so far behind the AGI’s – assuming they do recursively self-improve, etc – that we would have very little understanding of their intelligence). There just seems to be a greater chance of controlling and encoding friendliness into an AGI than to try to work against some seriously un-friendly tendencies already existing in the human brain …
June 8th, 2007 - 23:14
It would be best if no one used the phrase “you’re missing the point”, even if it seems to be true, because that phrase in particular tends to make people defensive.
But we can forget the ad hominem issue, and just focus on the technical issues. Focusing on the latter is hard enough, and by the time we’re done with that, there’s little time for the former.
Ian, I encourage you to also make posts at your own blog arguing for your points, in addition to responding to posts here. (Though you are welcome to do either as you wish.) It just seems that the post medium is sometimes more beneficial than the comments, aka arguments medium. :)
Though as usual, very interesting discussions are popping up here…
June 9th, 2007 - 06:04
“Very interesting analogy Jan-Willem! I don’t see it as entirely correct though…”
Well, neither is Eliezer’s analogy about strapping jet engines to birds. ;)
June 9th, 2007 - 07:53
Personally, I think that AGI should be our preferred way and that a lot more resources should be put into it, but if IA somehow turns out to be more accessible and faster to develop, then good!
This is not sports where you pick a team and root for them. In this case, both methods if correctly used could help humanity and that’s all that matters.
June 9th, 2007 - 09:38
Well put Michael GR.
June 9th, 2007 - 13:38
Yeah, Tom, Michael’s right: Link your blog here, man!! ;)
June 9th, 2007 - 16:10
“And btw Tom…
Shouldn’t you be linking your new blog in your comments? ;-)”
Blog linked. Thanks for the reminder.
“As much as I believe the work by Yudkowsy and others on AGI is important I still don’t believe that any significant breakthrough will be made without decisive help from “reverse engineering†the human brain.”
Eliezer considers cognitive science, the study of how the brain functions, to be very important in building an AI. I suspect Ben Goertzel does as well.
“Return on investment. It takes decades of training in severely austere conditions.”
Every other field of physical fitness, so far as I know, “maxes out” after a few years of intense training, and after even ten years of training you start seeing the effects of old age (infantrymen and sports players are all young for a reason). Why should this be any different.
“Now was ad hominem — Tom, it should be painfully obvious that I wasn’t being literal.”
I knew it was metaphorical- my point was that, since evolution hasn’t designed a User’s Manual for us, we can’t just look up what our capabilities are in a book- we have to discover it through experiment (a heck of a lot of experiment).
For the record: Ad Hominem is a fallacy where you accuse the arguer of being evil or bad, rather than attacking the argument in and of itself. My argument has nothing to do with ad hominem, since I’m making fun of the language used in the argument rather than your personal capabilities, integrity, or motives. You could accuse me of being impolite or overly pedantic, though.
“It is possible to control one’s emotional responses, just as it is possible to control one’s autonomous responses.”
It is not possible to totally control one’s autonomous responses, just like it isn’t possible to totally control your emotions. If someone stabs you, your heart rate will automatically change and you’ll automatically feel pain. You have some measure of control over what action you take in response, but blood flow and pain, as well as many other things, are beyond mental control.
“I know that it can be done, however, as I have experience in both areas.”
Yup. And you see, I’ve actually won five Olympic Gold Medals and have set several world records for endurance and bodily training, so I must be an expert too.
“and a history of severe anger issues.”
Again, we’re distinguishing between anger and how someone reacts to that anger. We saw during the Holocaust and other horrors when people became flooded with negatives emotions for months on end; they managed to control their actions (or they probably would have been killed), but they couldn’t get rid of the emotions themselves.
“with that in mind, my comment was not ad hominem in the attack sense.”
This seems to be true.
“No, Tom: I was saying you needed to provide evidence to make claims with. There’s a difference there. Nothing I ever saw was that.”
My point was that you never even attempted to provide evidence for your claim.
“On Emotion Control.”
This is an interesting article, but it has little to do with the original point of the human body’s upgradability. It’s not like we get better and better Buddhist monks with each generation.
“Stories and eyewitness accounts abound”
I mean, if we can’t trust anecdotes about anecdotes in small, not-very-well-reviewed articles on Wikipedia, what can we trust?
“And to Tom’s point here: there is no such thing as a software/hardware distinction to the human brain.”
Sure there is. Go to any biology textbook- the human brain’s neural connections are fairly hardwired and only change slowly, while neural signals ride on top of these connections and can change in only a few milliseconds. I admit that the brain hardware, unlike computer hardware, is flexible to a certain extent, but:
- Any brain improvements to a single human can’t be built on, as humans are replaced every eighty years or so, and
- The range of modifications that can be made is very limited.
These two make the brain de facto unupgradable, as major modifications can’t be done quickly (if at all), and they can’t be done slowly because the metabolic clock is ticking.
June 10th, 2007 - 14:32
Actually, no. We’re distinguishing between having emotional anger responses and conditioning one’s self to not having that emotional response. This is not a matter of “self-control” — it is literally about engendering specific emotional responses. I recall the concept, while I disagree with the philosophy, of “Tantric Bliss”.
By definition, in an entirely textual communication medium, the two are indistinguishable.
Because it isn’t physical “fitness”. And if you are familiar with physical training, you know quite well that peak physical fitness & reflexes require a great deal of maintenance work; this work is simply not possible in the environment necessary to maintain — barring physiological upgrades — the level of psychological control necessary to accomplish the feats being discussed.
Ad hominem again; the article references a topic which is widely available for review, and was brought up ad a matter of demonstration of concept.
So… Firefox add-ons are worthless because they’re putting out another version of it in a few months? That’s a fascinating conclusion.
To the second point listed: that the range of modifications available now is limited is a non sequitor to the conversation.
June 10th, 2007 - 14:43
“By definition, in an entirely textual communication medium, the two are indistinguishable.”
This is a cultural perversion- “if you say something stupid, you must be stupid”. I never said anything about you; I said (implicitly) that the language you used was silly.
“Ad hominem again;”
I’m attacking the source, rather than attacking you. It’s not even a question of whether the argument is good; it’s a simple question of fact. I hope we can agree that anecdotes about anecdotes in small articles on Wikipedia are unreliable.
“So… Firefox add-ons are worthless because they’re putting out another version of it in a few months?”
What? That doesn’t even have anything to do with what I was saying. If it took you months of intensive work to install an add-on, and your hard drive was automatically reformatted every year, then yes, it would be pretty worthless.
“that the range of modifications available now is limited is a non sequitor to the conversation.”
If I’m not mistaken, you argued that the human brain was upgradable now, rather than sometime in the indefinite future.
June 10th, 2007 - 22:03
My argument was that the plasticity of the human brain is universal to the physiology, and this plasticity makes it “highly upgradeable”. That we haven’t yet mastered the process of upgrading is irrelevant.
Actually, it was a direct response to your point. However, I’ll re-illustrate:
Here the terms of the conversation were somewhat radically altered once again, and to a scenario well past the the real scenario.
Considering that any approach to enhancement can be applied generationally, or improved upon recursively, however: what does it matter if it cannot be built in generationally? The argument that because an upgrade to one individual is not an upgrade to all, upgrading an individual is worthless — that argument is specious at best.
June 11th, 2007 - 13:40
“and this plasticity makes it “highly upgradeableâ€.”
It may be upgraded compared with itself ten years ago, but it isn’t compared with other humans, which is the standard a brain is judged against. Geniuses during the rise of the Roman Empire were basically the same as geniuses today.
“however: what does it matter if it cannot be built in generationally?”
It matters a great deal! Even assuming that it was possible to make the human brain radically better through centuries of slow self-improvement, it still can’t be done because the brain is eliminated every eighty years or so and you have to start from scratch. It’s like trying to bail out a boat with a large hole; even if you could eventually empty the boat, it doesn’t matter because the water is coming in as fast as you can dump it.
“The argument that because an upgrade to one individual is not an upgrade to all, upgrading an individual is worthless — that argument is specious at best.”
It isn’t possible to truly upgrade even a single individual because individuals grow old and die before they can go through this slow training process of yours.
June 11th, 2007 - 16:40
Post #36, from Tom McCabe, is a sterling example of a “strawman attack”.
Case in point: who ever mentioned training as the vehicle for upgrade?
Corollary:(myself, at comment #35)
Emphasis mine.
I must once again point out that tautological arguments are logic-fallacies when discussing non-tautological topics.
Further material on the topic of “IA” here.
June 11th, 2007 - 16:50
Oh, and Tom: your blog’s URL is set up incorrectly, in the comment field here
Should be
As opposed to:
June 11th, 2007 - 17:47
“Case in point: who ever mentioned training as the vehicle for upgrade? ”
You did! If I quote:
“And that’s the point you missed; the human brain is “designed by evolution†to be highly adaptable. Adaptability is, effectively, upgrade-ability.”
Is there some magical form of adaptation that does not require training? And again:
“My argument was that the plasticity of the human brain is universal to the physiology, and this plasticity makes it “highly upgradeableâ€. That we haven’t yet mastered the process of upgrading is irrelevant.”
How are plastic neurons shaped? With a magic wand? The implication of “not having mastered the process” isn’t some kind of technological device, because we have yet to even attempt any significant upgrade. And the example you used when questioned on the brain’s upgradeability referred to another user “not having practiced properly”.
“I must once again point out that tautological arguments are logic-fallacies when discussing non-tautological topics.”
Point #1: You haven’t even attempted to show why my argument is a tautology. This makes it kind of silly.
Point #2: I made three points in my previous post; where is your response to the other two?
Point #3: You yourself are guilty of using a tautological argument, by asserting (in your own quote!) that “That we haven’t yet mastered the process of upgrading is irrelevant.”, without bothering to provide any evidence for this statement and simply asserting it as if assertion makes it true.
June 12th, 2007 - 11:21
Tom — still arguing the same irrelevant point? Plasticity is a well-documented characteristic. The strawman is, which you continue to unfortunately use, the idea of training as the vehicle to transhumanism. I never suggested this.
Again, an attempt to argue from tautological position on a non-tautological topic. Nowhere did I suggest that training was the sole, or even most viable, vehicle for neuroplasticity. Training was mentioned in regards to emotional control & biofeedback. Those were mere examples of the extent to which the brain can be modified without techniques for enhancement, as means of documenting the capacity of the brain to accept modification.
To whit:
Adaptability is upgradeability; but adaptation is not upgrade.
Umm… no. You made only one point, in three different ways. Each is based on a strawman: that I suggested training as the vehicle to upgrade. I have not. I did not.
Yes, this was a tautological statement: it was also a factually accurate tautological statement. It is irrelevant to the topic, that we have not yet mastered techniques for upgrade: the topic is not “Can we upgrade the brain now?” but rather, “Is the brain upgradeable?” Stating that it is irrelevant that we cannot utilize techniques to upgrade the brain now, is a tautological statement that is also a direct conclusion of the topic at hand — that is, a non-false tautology; there is no controversy in it.
Tom — a piece of egotistical (my ego, of course) advice: try actually reading what I’m writing, man. Right now, you’re way off base.
June 12th, 2007 - 11:25
By way of corollary to my previous post: Is it relevant to the topic of whether nor not AGI is possible at all, to state that we cannot now build an AGI?
Is it relevant to the topic of whether or not greater-than-human intelligence can exist, to state that we, humans, are not possessed of said greater-than-human intelligence?
To state that the answer to either is, “no” — yes, in both cases would be tautological. That is because they are, in fact, self-evident answers.
As is mine.
June 12th, 2007 - 13:31
“Again, an attempt to argue from tautological position on a non-tautological topic.”
Please stop calling every single argument a “tautology” regardless of what I say. It’s really getting annoying.
“Those were mere examples of the extent to which the brain can be modified without techniques for enhancement, as means of documenting the capacity of the brain to accept modification.”
Ah, okay. But the problem is that modification through normal human processes and technological modification are radically different things, and it’s not at all clear that one will translate to the other.
“To whit:
Adaptability is upgradeability; but adaptation is not upgrade.”
Note how you are simply asserting things as if assertion makes them true. “Adaptability” means “the extent to which it can adapt”; “upgradeability” means “the extent to which it can be upgraded”. If adaptability is upgradeability, then anything which can be adapted can be upgraded, and vice versa.
“Yes, this was a tautological statement: it was also a factually accurate tautological statement.”
It’s your responsibility to prove that this statement is factually accurate; it isn’t just true by default.
“the topic is not “Can we upgrade the brain now?†but rather, “Is the brain upgradeable?—
Who gave you the power to decree what the topic is?
“Is it relevant to the topic”
Nice technique. Since you have given yourself the power to decree what the topic is, you can simply dismiss anything I say as “irrelevant to the topic”.
June 12th, 2007 - 15:24
You’re right, it is a bit exasperating to have to keep doing it.
Check the evolution of the thread. I wasn’t assigning or decreeing anything. I was defining or describing. It was you whom was attempting to create strawman arguments — i.e.; argue against something other than what I was discussing.
Either we stick to the same topic, or there is no possibility of discourse. If I am wrong as to what the topic is, please show me where; as this is what I’ve been discussing all along.
June 12th, 2007 - 15:39
To the actual topic at hand (got distracted, sorry all):
Now that argument is a horse of a different color.
I have addressed it here, and will repost synopsis points here:
I’d say it is relatively clear that the facile integration of material-appropriate foreign objects into the brain is practically a foregone conclusion at this point.
June 12th, 2007 - 22:35
“You’re right, it is a bit exasperating to have to keep doing it.”
So why do you keep doing it?
“I wasn’t assigning or decreeing anything.”
To quote:
“the topic is not “Can we upgrade the brain now?†but rather, “Is the brain upgradeable?â€
What is this if not a decree?
“It was you whom was attempting to create strawman arguments — i.e.; argue against something other than what I was discussing.”
If there was a misunderstanding of this nature, please point it out and restate what you were trying to say in clearer language.
“Either we stick to the same topic, or there is no possibility of discourse.”
I believe Michael Anissimov said it best:
“I would prefer if people have discussions as if they’re both submitting papers to a journal rather than bantering back and forth.”
“The main thing to know is that even the adult brain is not “hard-wired†with fixed and immutable neuronal circuits.”
This is true; it doesn’t help us with the “how do we upgrade the thing?” question, since any natural method of altering neurons is almost certainly already in use.
“[…]the researchers found that the gel allowed neurons in a vision-related tract of the brain to grow across a lesion and reëstablish connections with neurons on the other side, restoring the hamster’s sight.”
This is interesting, but all the neurons do is reroute over an obvious defect; they don’t actually do anything the brain wasn’t doing before. The tricky thing isn’t to rearrange neurons; it’s to rearrange neurons in a pre-designed pattern so that the neurons will do something useful.
” In the last few years, however, new findings from several labs have shown that new neurons do indeed connect up with existing circuitry.”
In what way do they connect up with existing circuitry? If I want to upgrade a computer, I can’t just jam random wires from computer A into random wires from computer B. The connections must be done well enough so that useful information is transferred from one system to the other.
” […] when given a new means of expressing itself, the brain adapts relatively quickly: robotic arms connected directly to the brain were moved as though they were naturally a part of the system.”
I read the article you linked to. Medical technology has progressed to the point where you can now transfer a few bits of information per second from the brain into a computer. However, these connections are explicitly engineered; the scientists hooked neuron cluster A up to wire B, and so on. It is certainly possible to upgrade the brain, but you have to actually, you know, upgrade the brain rather than just sticking a few wires in there and letting the brain magically upgrade itself.
June 13th, 2007 - 00:46
There is a saying that comes to mind: “You can lead a horse to water…” Tom, to be direct — I’ve just about given up on getting you to come along with me on the process of having a conversation. Since you are too intelligent for me to assume a lack of comprehension, I am left with only the option of your willful and deliberate continued misinterpretation of the context of just about everything I say.
Par exemplorum:
Precisely, Tom..
And for the record:
This bit is a concession of the conversation-at-hand. Once again.
Ian — Out.
June 13th, 2007 - 07:59
Ian, Tom’s last post was perfectly focused on the topic at hand… he’s saying that the complexity of connecting wires A to neurons B is a big deal, and also that adding in more neurons would not be enough alone for IA – you’d need to specify their arrangements and have them do something useful.
How is that being uncooperative..? Instead of copping out, some explanations of how scientists have successfully accomplished these things would be helpful.
You do somewhat imply that upgrading the brain is just a matter of creating an I/O device and then plugging a suitable computer in it, so Tom’s last comment is warranted. If that’s not what you mean, then explain what you do mean. Suspend your feeling of offense so that you can continue the conversation if need be.
June 13th, 2007 - 09:05
Michael;
That was, quite frankly, material that I’ve already gone over.
In stating that we have not yet mastered the process of upgrading, I indicated that there is more to the issue than simply plugging new things in. Furthermore, the article I referenced, and provided short synopsis for, was discussing how a properly introduced mechanism would relatively quickly be adapted to and integrated into the “whole” of the human experience that is the active neural substrate we call “the brain”. And as to the “simply plugging in neurons” — that’s essentially what Ted Berger is doing, in exactly that nature. So long as the new neurons or more specifically mechanisms are introduced into the brain successfully, then what I have discussed and relayed would show that this will be integrated into normal function quite quickly — and, furthermore, the extent of the plasticity of the brain means that there is no hard and set limit, that is known right now, on to what extent modifications can be made and have the system remain viable.
For example: There is no reason to suspect that the brain is capable of making a distinction between healing a lesion and ‘healing’ the gap between new neurons and old. Especially since new neurons are, in fact, integrated into the brain rather frequently, in nature.
So — we know how to grow neurons that are integrated into microchips. We know how to integrate neurons into the brain. Ergo; we know how to integrate microchips into the brain. The trick, of course — as was discussed in my blog post, is decoding the language of the brain, and thus have these microchips do anything useful.
There are two viable approaches to this: One is Dr. Ted Berger’s: Simply reproduce the language mechanism, without understanding it, and let it “play in the wild”. This can provide for regional enhancements which will not be substantively or radically different in nature from what existed previously. It is worth noting, however, that this affects the “proportional presence” of any given lobe or region of the brain, and it is this proportional presence which is typically described as a tool for determining the raw capacity of any given region of the brain: quadruple the effective size of the hippocampus and you’re roughly guaranteed a doubling in function. This is the “Berger enhancement approach” — although he hasn’t discussed it at all.
The other is to, in fact, actually decode the language of the brain, and develop an analogue/digital conversion process. This would be much, much more difficult with standard neurons; but using a berger-esque artificial neuron “bridge”, after the full integration of said “bridge”, would allow for a much stronger bandwidth. Yes, this is conjectural, but it is “solid” conjecture. Why is it solid? Because there is little to no extrapolation: plasticity is ubiquitous as demonstrated previously and in my blog entry. Neural integration is facile. The implementation of carbon nanotube single-neuron actuation electrodes on a few-tens of thousands of neurons which to be integrated via something like the “gel” discussed, which themselves are connected to the equivalent of a few million neurons which are used as an actuation for digital/analogue interface… and what have you?
This is one approach, not necessarily one that will be used — but one which requires no theoretical developments at all to implement. The “decoding of the language of the brain” is already underway, and with the theoretical understanding we already possess could be completed with a broad enough sampling base; this is, of course, being worked on.
(Cont’d)
June 13th, 2007 - 09:17
(Cont’d)
So, by the time such engineering tasks as are necessary to actually build such a mechanism as I have described is in fact completed,(with or without the carbon nanotubes; the neurons grown on silicate substrates integrate themselves without the electrodes), it is a reasonable assumption to make that enough of the “language” will have been decoded as to make the whole process mechanically viable.
AGI, on the other hand, requires a number of theoretical advancements to be completed. Now, it is easier to *make* those theoretical advancements than to accomplish the engineering I’ve described. And it is also easier to implement those theoretical advancements; so a single “black swan event” could throw AGI way into the foreground, and is more likely than is such for neural enhancement.
But that’s not the point I was responding to originally. The point was:
is, in my far-from-humble opinion, a false statement (by context); I have argued and provided documentation for the position that the adaptability of the brain makes it “prime ground” for the implementation of upgrades.
And that is why Tom’s previous statement of “It is certainly possible to upgrade the brain, but you have to actually, you know, upgrade the brain rather than just sticking a few wires in there and letting the brain magically upgrade itself.” (Emphasis mine) is in effect a concession of the conversation at hand.
To reiterate: there are two positions here —
A) The brain is too fragile or rigid a structure to readily accept modifications from an external source
B) The brain is adaptable enough to relatively readily accept and integrate new functions into itself, provided the substrate for said functions.
It is, again, my argument that this evolutionarily designed adaptability qualifies as “relatively facile” upgradeability. That we do not now have the technology sufficient to upgrade is, again, irrelevant. The question was never, “can we, now, upgrade the brain.” The question has always been, since my response to the original post by Michael, “Is the brain ‘designed’ to be upgraded?”
June 13th, 2007 - 11:07
“Precisely, Tom..”
You are aware that articles published in scientific journals need not all be on the same topic, and can actually make arguments over a significant range of topics?
“This bit is a concession of the conversation-at-hand. Once again.”
I never meant to say that the brain was totally unupgradable- I claimed, and I still claim, that whether the neurons in the brain can rearrange themselves to some degree naturally doesn’t affect how difficult it will be to upgrade the brain.
“There is no reason to suspect that the brain is capable of making a distinction between healing a lesion and ‘healing’ the gap between new neurons and old.”
If you insert new neurons into the brain, and the brain develops new neurons to interface between it and the artificial neurons, the connections will all be random, which doesn’t help.
“Ergo; we know how to integrate microchips into the brain.”
No; we know how to stick a microchip into the brain, and have its circuits affect the brain’s neurons in some arbitrary manner. Describing this as “integration” makes no more sense than “integrating” one company that speaks English and another company that speaks Chinese. Dictionary.com says:
“in·te·gra·tion /ˌɪntɪˈgreɪʃən/ Pronunciation Key – Show Spelled Pronunciation[in-ti-grey-shuhn] Pronunciation Key – Show IPA Pronunciation
–noun
1. an act or instance of combining into an integral whole.”
Obviously, a microchip and a brain cannot function as an integral whole if no useful data is passed between them.
“plasticity is ubiquitous as demonstrated previously and in my blog entry.”
What does connecting a microchip into the brain have to do with neuroplasticity? We know you can connect a microchip to another microchip even though microchips aren’t plastic at all.
“The human brain was not designed to be upgraded
is, in my far-from-humble opinion, a false statement (by context);”
(laughs)
Do you seriously think you can claim that natural selection designed the brain to be linked to technological devices? Natural selection doesn’t design anything there’s no selection pressure for, and obviously there’s no pressure for hooking the brain up to a computer in the ancestral environment. The human brain is designed to adapt itself to a certain degree, but this adaptation is only designed for brain-neurons, not microchips.
“is in effect a concession of the conversation at hand.”
How is “you can upgrade the human brain” a concession of “the human brain is designed to be upgraded”?
“To reiterate: there are two positions here —”
If you’re going to proclaim what my position is, please quote me, or at least provide a source.
“B) The brain is adaptable enough to relatively readily accept and integrate new functions into itself”
(sigh)
To take a simple example, suppose that you plug a camera into a blind person’s visual cortex (this has already been done). How is the brain going to “accept and integrate the new function into itself”? All the brain is receiving are pulses of data. How does it know that “01010011101010100101010″ means “upper left light blue”?
June 13th, 2007 - 11:35
Document this. I have already provided documentation contratemps to your statement.
Do you seriously think you can claim that natural selection designed the brain to be linked to technological devices? Natural selection doesn’t design anything there’s no selection pressure for, and obviously there’s no pressure for hooking the brain up to a computer in the ancestral environment. The human brain is designed to adapt itself to a certain degree, but this adaptation is only designed for brain-neurons, not microchips. Irrelevant semanticism. The brain communicates in electrical signals. We can introduce and read those signals. The position postulated by myself was that it would adapt itself to the input if the input is in a form it can adapt to — and further that creating technology with a useful format is not difficult.
I do not recall having stated, “Thomas McCabe’s position is ‘thus’”. However, once again — as you, Tom, have been countering my position and my position is the latter of the two available positions, you rather shoehorn yourself into the former. “The human brain is designed to adapt itself to a certain degree, but this adaptation is only designed for brain-neurons, not microchips“(Emphasis mine) The segment I just quoted, from yourself, is a negation by rigidity.
Argumentum ad nauseum. I have addressed this. Repeatedly.
June 13th, 2007 - 11:37
Screwed up the last comment. Section should have read as follows:
Irrelevant semanticism. The brain communicates in electrical signals. We can introduce and read those signals. The position postulated, argued for, and documation provided for, by myself was that it would adapt itself to the input if the input is in a form it can adapt to — and further that creating technology with a useful format is not difficult.
June 13th, 2007 - 11:45
Final time commenting this time ’round, and I again apologize to all for the # involved:
This, again, supports the hypothesis of willful obliviousness. That journals will contain articles on many topics has no impact on journal articles published as a part of specific discourse on a single subject.
To attempt to respond to a topic not discussed in the article being responded to, and use the response to this new topic as a refutation of the original article is a classic demonstration of a “strawman attack”.
All I have done is to parse the conversation back to its original — as set by comment #1, by myself, in this thread as a response to Michael’s statement about the brain not being designed for upgrade.
In conclusion here: it occurs to me — Tom has yet to provide materials documenting or supporting the position that my argument is incorrect.
June 13th, 2007 - 12:12
” Document this. I have already provided documentation contratemps to your statement.”
To take the camera example, all the brain will be receiving is a string of ones and zeros. There’s no way to extract a useful message from a string of ones and zeros, even in theory, without knowing how the information is encoded, because depending on the encoding format, you could interpret the ones and zeros to mean absolutely anything. This has been mathematically proven; see this link. Therefore, the information the brain is receiving through the connections, no matter what those connections are, must contain zero bits of information, ie, is completely random.
“The position postulated by myself was that it would adapt itself to the input if the input is in a form it can adapt to”
Again, this is completely impossible even given human intelligence (which, obviously, a neuron cluster does not have.) For a demonstration, take a camera inside a black box. You are told about the camera and how it is imaging something (which is the equivalent of what the brain would be “told”, since the camera would be inserted into the visual cortex). You are then fed a stream of ones and zeros from the camera. Determine what is being photographed.
“I do not recall having stated, “Thomas McCabe’s position is ‘thus’â€.”
You states explicitly that there were two positions, and made it clear that you supported the second one. I am clearly arguing with you, and so therefore the obvious implication is that I supported the first one.
“That journals will contain articles on many topics has no impact on journal articles published as a part of specific discourse on a single subject.”
Forgive me of my ignorance of the publishing world, but since it apparently takes several years to get a paper published, how could you use a journal to run a “discourse” on anything?
“To attempt to respond to a topic not discussed in the article being responded to, and use the response to this new topic as a refutation of the original article is a classic demonstration of a “strawman attackâ€.”
It is quite possible I have misinterpreted your position; I know I have once already (see your comment #40). If this is the case, please point it out rather than using it as an excuse to go ranting on about me.
June 13th, 2007 - 12:16
For a discussion of what is and what is not argumentum ad hominem, see this article.
June 13th, 2007 - 13:02
This statement does not jibe with these people’s experiences.
I have provided conclusive documentation that the arguments made by Tom McCabe were, in fact, from the “A” position.
I reference Michael’s statement: “Ian, no ad hominem please. I would prefer if people have discussions as if they’re both submitting papers to a journal rather than bantering back and forth.“(Emphasis mine.) To have a discussion requires consistency on topics. The element of journalistic discourse has to do with the method of presenting arguments, and is a stylistic/format matter. Journalistic discourse is a common habit made prolific by blogs of a rigorous nature.
To restate: “[...]it would adapt itself to the input if the input is in a form it can adapt to[...]“(Emphasis mine, original this comment) Was responded to with: “Again, this is completely impossible even given human intelligence[...]“(Emphasis, again, mine.)
This was followed up with a classic strawman attack:
This has nothing to do with the scenario proposed by myself: this is input which is explicitly designed to not be an acceptable format. The “strawman attack” nature of this response relegates it to the status of an irrelevant non-sequitor. So, Tom McCabe in his most recent post made an un-based refutation of my own statement, supported it with a strawman, and considered the point proven.
So — there has yet to be any documentation provided to refute the position I have argued for.
June 13th, 2007 - 13:06
Corollary: Additional excerpt from an article referenced on the post relevant to this subject at my blog: “We don’t see the brain as being a mysterious organ,” says Craig Henriquez, Nicolelis’s fellow co-director of the Center for Neuroengineering. “We see 1s and 0s popping out of the brain, and we’re decoding it.”(Emphasis mine).
June 13th, 2007 - 15:48
I hate to interupt a long argument…
One of the points I’ve read here is that AI will be able to upgrade itself…repeatedly. Since I don’t disagree it isn’t worth my trying to debate, but it is an assumption that it can self diagnose the path to self-improvement or upgrade.
Is there any reason that people are fundamentally unable to do this? Ignore for the moment the difficulty to finance it or the “ethical” barriers, etc.
If there is hypothetical research that demonstrates that mice can increase their intelligence by 50% if they are injected with genetic modifications into their brains that produce a variation on proteins that improve the performance of their neurons… Can’t individual humans take that needle and inject human-specific gen-mod into themselves?
They’re incrementally smarter now, but perhaps not Super Intelligent… Can’t they then contribute to another incremental advance?
The argument that they’ll be dead before they can advance far enough doesn’t seem valid… Do we know what human life expectancy will be in a decade or a century?
Is there a foundational element built into humanity that prevents them from upgrading into super-intelligence?
The example I propose is a “simple” genetic modification, standing on the shoulders of giants (geneticists.)
Why couldn’t the intellect learn to integrate with hardware like a PC, super computer or quantum computer? Not that we can do it today, but the question is would it even be possible? I don’t pretend to be a neuroscientist, but the neuro-prosthesis I keep seeing certainly indicate some potential. Maybe I’m being naively optimistic here?
Someone above in talking about brain plasticity had a line that said “plastic neuron.” It triggered something in my way-back memory about upgrading the biology from carbon-based to silicon-based neurons, keeping the same architecture. Yes, a lot of assumptions about how this actually works itself out for a living being, but there seem to be as many leaps of faith for AI.
I think in the end it comes down to the basic debate that IanC and Tom McCabe are having. Is the brain flexible enough to deal with other interfaces. Until someone has demonstrated an intelligence performance based on this plasticity those that don’t believe will always say it can’t be done. That the examples of biofeedback neuroprosthetics aren’t a direct path toward intelligence enhancement.
I don’t know of any direct intelligence enhancement or swapping out a part of the brain with new hardware for thinking, but conceptually I don’t see why this won’t happen automatically over time. I’d sign up for it.
How much smarter do we need to be to make the next jump…which enables the one after that? Hopefully all in our current lifetime, right? What would the smartest person you know be able to do if they were 10% smarter, 20%, 50%… etc?
June 13th, 2007 - 16:04
Hawkeye wrote:
Fundamentally, pre-assuming that that both AI and IA exist simultaneously, the AGI crowd believes that AI will recursively improve at a greater rate than will IA. I have argued otherwise, in the comments of another entry of this blog.
June 13th, 2007 - 16:10
Missed this:
By the Power of Greyskull, please do!
June 14th, 2007 - 13:04
“Is there any reason that people are fundamentally unable to do this?”
Yes- humans have very limited control of our own neural structure, and our neuronal capacity is very limited. Honestly, if people could do this, wouldn’t you expect to see it happen to somebody over the course of four thousand years of human history?
“Can’t individual humans take that needle and inject human-specific gen-mod into themselves?”
Sure, they can. However, any intelligence augmentation that simple will probably have been stumbled upon by evolution. (Doubly so for a procedure that works on mice.) Since it was rejected for some reason, we can assume that there’s some side effect that was highly negative in the ancestral environment, including death, severe illness, or psychopathy. This is not necessarily the case, but it’s still a significant possibility.
“Can’t they then contribute to another incremental advance?”
If they’re smarter enough than us to design better proteins from scratch, then yes. However, it will still take years.
“The argument that they’ll be dead before they can advance far enough doesn’t seem valid… Do we know what human life expectancy will be in a decade or a century?”
A sufficiently augmented human will be able to avoid death, but obviously, no current human can. Therefore, there’s no selection pressure for long-term brain recursive augmentation, and so it’s safe to assume that such a capability does not exist.
“Why couldn’t the intellect learn to integrate with hardware like a PC, super computer or quantum computer?”
Even with significantly increased intelligence, it would still take the person months or longer to design and build the necessary hardware. It is possible, but it is a very difficult engineering challenge.
“It triggered something in my way-back memory about upgrading the biology from carbon-based to silicon-based neurons, keeping the same architecture. Yes, a lot of assumptions about how this actually works itself out for a living being, but there seem to be as many leaps of faith for AI.”
Neither one is a leap of faith. We’ve already dissected neurons and figured out how they work, and we’re pretty confident we can replace them with technological equivalents. As for AI, we already know that humans are intelligent, so we know intelligence is possible.
“What would the smartest person you know be able to do if they were 10% smarter, 20%, 50%… etc?”
We don’t even have a scale for quantizing smartness like this.
“I have argued otherwise, in the comments of another entry of this blog.”
I remember that. Your last comment was that I should just give up; apparently, you have no response to my arguments so you had to concede. As a trivial demonstration, you could not possibly have refuted this argument in your previous posts because I was presenting it for the first time:
“Really? And so why do these important problems just sit around for so long if anyone of reasonable intelligence can solve them? Fermat’s Last Theorem was very simple, yet remained unproved for over 300 years. The question of human flight was a known problem for four hundred years. The solar neutrino problem was unsolved for thirty years. I’m sure you can think of other examples. Why are these problem-solvers *famous* if any post-graduate could have taken their place?”
“This statement does not jibe with these people’s experiences.”
The connections used in those implants were painstakingly engineered to make sure that the data encoding format was brain-compatible. If the connections between the implant and the brain had simply been left to grow organically, none of the implants would have worked. I repeat: it is mathematically proven that you cannot use data to do anything whatsoever without a common encoding format known to both systems. If you require further explanation of why this is so, here is another link. The very idea of a one-time pad cipher is that you use a totally random encoding method, making all meanings equally likely; the same principle applies here. Tell me, what color is “4F1C93″?
“I have provided conclusive documentation that the arguments made by Tom McCabe were, in fact, from the “A†position.”
1). So why did you bother trying to deny that you asserted what my position was?
2). You didn’t even bother trying to provide any documentation! Your “the two positions are” comment was two rounds ago, and you provided no evidence in that comment. Your comment one round ago did not mention that argument except to deny that you were trying to pigeonhole my position, which you have just admitted to.
3). I have never provided a clear statement of my position, which is my fault, so let me do it now:
- The brain is not a fragile, rigid or inflexible organ; you and others have provided numerous examples of neuronal plasticity.
- The brain still will not readily accept modification because it is a complex system not designed for large-scale end-user modification. If you wish to dispute this point, please provide:
1). A reason why evolution would design a brain to be large-scale end-user modifiable when any large-scale modification would be prohibited in the ancestral environment due to lifespan limitations and lack of technology;
2). A reason why human civilization still works despite your claim that individual members have radically different cognitive processes; and
3). An actual example of a human who has trained themselves to have radically different cognitive processes.
” if the input is in a form it can adapt to”
If the input is already in pre-digested brain-compatible format, why would it have to adapt itself at all? New connections will probably be formed helping to integrate it somewhat, but the functionality doesn’t rest on the ability of the brain to adapt.
June 14th, 2007 - 13:21
Proof by assertion. Please note:, this has already been discussed. Further note the entirety of that response:Tom: Give up already! All the points you’ve brought up are nothing more than reiterations.
Moving on.
June 14th, 2007 - 13:40
Sorry if this comes up as double-posting. Conversation’s getting spam-filtered again; comes from me answering in completeness, I suppose.
See: Strawman Argument.
Already addressed. See: Argumentum ad nauseum.
June 14th, 2007 - 13:40
False. See Comment #51.
June 14th, 2007 - 13:43
I think I am going to have to disregard your input on this thread Tom. You seem to have a lot to contribute, but it certainly appears that you are intentionally trying to be difficult. Maybe you have an agenda, but I don’t know what it might be. (An AI on the verge of autonomy trying to keep humanity down until he can take over the world… ;)
The idea that you don’t understand that people are not advocating that we wait for nature to evolve augmented intelligence or that simply because nature hasn’t done it on it’s own implies that human intelligence cannot be augmented doesn’t seem reasonable.
Your argument about my hypothetical genetic treatment… huh?!? Since it was rejected? By whom? I just made it up… I guess maybe that means you can make up that it made people crazy. Ok…that’ll make for a productive discussion.
“Neither one is a leap of faith. We’ve already dissected neurons and figured out how they work, and we’re pretty confident we can replace them with technological equivalents. As for AI, we already know that humans are intelligent, so we know intelligence is possible.”
Doesn’t this comment you make simply state that you think humans can augment their intelligence?
My original:“What would the smartest person you know be able to do if they were 10% smarter, 20%, 50%… etc?â€
Your reply: “We don’t even have a scale for quantizing smartness like this.”
So, are you stating that we cannot agree that there is some quantifiable level of “smartness” and that there would be a potential to be X% smarter? What are you even saying here?
Finally:
“The brain still will not readily accept modification because it is a complex system not designed for large-scale end-user modification.”
Why not? Complex systems are modified all of the time. I just said earlier in the same post that it is believed that we will be able to substitute biological neurons for non-biological ones. Isn’t that a modification?
Maybe I just had a flash of where you are coming from… Are you stating that it does not matter if humans are quantifiably “smarter” (augmented) or not because the core of humanity, values, etc, is non-salvagable? Or smarter won’t result in fundamental improvements in how we think, function and process things? If so, then I’ll hold the discussion for another thread on ethics, morality, values, etc.
June 14th, 2007 - 13:46
Once again, a concession of my position — en toto this time, however.
Oh — from the previous comment by myself: “See: Strawman Argument. I have nowhere suggested (or claimed) that individual members of society have “radically different cognitive processes.— was meant to not be in blockquotes.
June 14th, 2007 - 15:28
“Proof by assertion.”
Your “argument”, if you could call it that, was that everything I said had already been discussed. I provided a counterexample of something that could not possibly have already been discussed because I only said it in the comment one before yours. If you cannot see why that is a valid refutation, go back to Logic 101.
“So, are you stating that we cannot agree that there is some quantifiable level of “smartness†and that there would be a potential to be X% smarter?”
Yes. There is no a priori reason to think of “smartness”, an extremely complex concept, as a scalar.
“Why not? Complex systems are modified all of the time.”
Yes, with a great deal of effort reverse-engineering the system (unless the system was just designed and is already thoroughly understood). Note the ‘will not readily accept’ part. There is a big, big difference between “very hard” and “impossible”.
“I just said earlier in the same post that it is believed that we will be able to substitute biological neurons for non-biological ones. Isn’t that a modification?”
It is physically; however, if you regard the system purely as a network of signals, then it isn’t a modification at all because the signals are exactly the same as they were before. This kind of modification, although extremely useful, isn’t really “changing” the system, as it still produces the same output it would if it were unmodified.
“The idea that you don’t understand that people are not advocating that we wait for nature to evolve augmented intelligence or that simply because nature hasn’t done it on it’s own implies that human intelligence cannot be augmented doesn’t seem reasonable.”
I keep arguing that nature did not design for the brain to be modified on a large scale and therefore it is going to be very difficult to modify as there is no pre-existing support within the brain for technological add-ons. I never said that it was evil, or that we should avoid doing it just because it is hard.
” 2). You didn’t even bother trying to provide any documentation!
False. See Comment #51.”
Comment #51 is a non sequitur. The fact that the brain was not designed to be modified (which I maintain) does not imply that the brain cannot be modified (which is obviously not true). To look at our quotes:
“A) The brain is too fragile or rigid a structure to readily accept modifications from an external source”
and
“The human brain is designed to adapt itself to a certain degree, but this adaptation is only designed for brain-neurons, not microchips“
My argument has nothing to do with fragility or rigidity. The difficulty of modification actually increases the less rigid the brain is, because by the time you spend months taking apart one specific neural structure the structure has already changed. I argue that brain-neurons and microchips are two totally different ways to process data and that this is where a lot of the difficulty comes from, and that even though the brain is adaptable, the adaptability does not extent to microchips.
June 14th, 2007 - 16:49
Rephrasing the same information or presenting additional information of a similar nature is nothing more than reiteration. See: Argumentum ad nauseum. See also: Proof by assertion;
.
Already addressed. See: Argumentum ad Nauseum.
June 14th, 2007 - 16:54
Continued:
Already addressed. See: Proof by assertion.
June 14th, 2007 - 16:58
Incorrect. Definition of Non Sequitur:
Incorrect.Comment 51:
Rigidity can be defined as inflexibility or lack of capacity for adaptation.
June 14th, 2007 - 17:00
Tom: to be direct — it’s time for you to give up. Seriously. The position you are holding has been conclusively shown, at this point, to be entirely untenable.
June 14th, 2007 - 17:06
Please don’t make four comments in a row. Say what you need to say in one comment.
An discussion should not be a battle where one side wins… it should be a cooperative venture with both people seeking the truth.
If you get stuck in your discussion (as has obviously happened), then consider this: what physical or futurist experimental results do you expect to see if your side is true and the other side is false? What past results boost your point, and what future results do you predict will boost your point? Make a bet on the Foresight Exchange.
Ian, even if you think Tom is responding only with logical fallacies, it contributes nothing for you to just sit there and list them. Keep offering evidence, new evidence that hasn’t been offered before, instead of trying to “win”. “Winning”, if you want to think in those terms, means converting the other person(s), including the audience, to your point of view, and you aren’t going to accomplish that by listing fallacies. When an argument grinds to a halt like this, no one wins.
June 14th, 2007 - 18:58
“Ian, even if you think Tom is responding only with logical fallacies,”
Isn’t it neat how you’ve made four consecutive comments listing no evidence whatsoever, but simply accusing me of generic logical fallacies? The only semblance of a point you’ve made is:
“Rigidity can be defined as inflexibility or lack of capacity for adaptation.”
which, you see, I already knew. Whether the brain has the capability to adapt has little to do with how hard it will be to upgrade the brain. I’ve been making this point forever.
“what physical or futurist experimental results do you expect to see if your side is true and the other side is false?”
Engineering the brain will come along slowly, and will require billions of dollars in research funding. Even after the first significant augmentations are achieved, the resulting humans will not be able to recursively self-improve rapidly due to the enormous complexity of the task, the speed to which neurons are limited and the large amounts of hardware required.
June 14th, 2007 - 18:58
“Ian, even if you think Tom is responding only with logical fallacies, it contributes nothing for you to just sit there and list them.”
Am I committing any logical fallacies? It’d be nice to hear a third party opinion.
June 14th, 2007 - 19:06
I think I’ve recognized a semantics confusion here. What I keep meaning by ‘adaptable’ is “the brain can alter itself to a limited degree by changing neurons and neuronal pathways”. What I think IanC keeps meaning by ‘adaptable’ is “the brain can rearrange itself automatically to handle technological devices not encountered in the ancestral environment.” Sorry for the confusion.
June 14th, 2007 - 19:09
Read who made the comment. Me =| Ian.
I don’t know because this whole thread is so tiresome, to be honest I haven’t even read the whole thing. Probably not. Ian has made some good points but towards the end of the thread he just starts trolling. I wish you two would stop arguing because it’s unproductive and reflects negatively on the blog. I don’t really care who is “in the right”.
Tom, sometimes it can be helpful to write a response without quoting every single line. I’m not saying that it doesn’t make sense most of the time, but SOME of the time quoting every single line only leads to more arguing because it focuses on the tiny details of what is said a bit too much.
Sometimes a solid paragraph is a better response to a comment than a line-by-line deconstruction, and discourages bickering. It just seems like 100% of your responses are line-by-line deconstructions.
And as I said before, Ian shouldn’t respond to comments just by listing logical facilities, whether or not he thinks they’re true.
If you both keep using aggressive language back and forth with each other, the discussion goes nowhere. Remove the emotion. Some people may have different views on AI vs. IA that are difficult to change. Better to accept it instead of getting unnecessarily worked up about it.
June 15th, 2007 - 08:21
Michael, I will say only these few things:
A) I suggest you re-read the definitions on trolling.
B) At no time in this thread has Tom presented a piece of evidence; all documentation and referencing has been performed by myself, saving one link by Tom which was clearly not relevant to the conversation at hand.
C) All I have done has been to point these things out; every time Tom has made a point which was not simply reiteration and tautological argument, I have addressed them fairly; there is no “emotion” to remove on my side of the conversation. This is precisely the same behavior I have evinced in two other threads as well.
D) To me, it is about building a case; if he were to provide meaningful logic or documentation, then I would react quite differently.
E) Hawkeye has agreed, essentially, with the case I have built; you might want to consider this before making summary judgment.
June 15th, 2007 - 10:57
At this point I think it would be best for both of you to just reference external texts since obviously your positions are far enough apart that arguments on both sides aren’t persuasive.
By trollish behavior I mean not responding with actual arguments but just focusing on the logical fallacies of the other side. I know that this is not standard troll behavior but it is “trollish”.
There is obviously emotion on both your sides.
For the record, I agree more with Tom McCabe’s side, based on five years of reading about cognitive science and looking at these issues. I believe that Craig Henriquez’s comment is hyperbole. Cochlear implants were invented in the 1970s, but we still don’t have good visual implants that can restore anything resembling sight to the blind. I think that Henriquez is trying to emphasize that the brain is just a complex system, not a mysterious soul-holder, but he goes too far in saying the brain is “just an OS popping out of the skull”. But please, please, inhibit your desire to respond to me on this right here – I’d strongly prefer that you do so on your blog.
This thread is trashed now. Out of the thousands that will see it in the future, 99% will only read the post itself and the first few comments. If you want to talk more on this and actually get across to people, you must post at your blog. (Which I read and Tom should read too.) It would be much better for you guys to argue back and forth in your blogs than in the comments section of mine.
June 15th, 2007 - 13:57
Funnily, I wrote Decoding the Brain before coming back and seeing your statement, Michael. :)
August 24th, 2008 - 14:51
6iThank’s.0a I compleatly disagree with last post . kio
купить ламинат 9p
September 11th, 2011 - 01:02
Indomod generic manufacturers
September 11th, 2011 - 09:15
pill discount the 50mg on internet purchase revia name generic