Lex and Kevin,

Thanks for the thoughts. Interesting dialectic. Kevin is in a big hurry. Lex says urgency doesn't help if you don't understand the basics. Kevin shows obvious sci-fi paranoia syndrome; big bad things with terribly simplistic ideas and limitless access to data will take over everything and kill or enslave everyone. Lex places inordinate almost spiritual praise on human ability and capacity.

Kevin, in particular you seem always to think of the present state of a system to be definitive for that system. You apply sweeping definitions to computing based entirely it would seem on the current incarnation. Computing is really only about 80 years old. I have trouble accepting that the current state of computing or even for that matter, the arch that has defined its first 80 years, in vary many ways defines the full topological or causal scope that is possible or definitive.

Also, it is important to remember that the boundaries we so thoughtlessly draw around entities (me, humans, culture, computing, my computer, etc.) though colloquially practical, are not meaningful when discussing energy use or efficiency. For instance, the existence of a computer is impossible except as a cultural extension or product of the activity of human beings. Human beings likewise are only possible as a result of billions of years of energy expenditure towards the gradual buildup of complexity that led from complex chemistry to biology to sentience. Chemistry itself owes its attributes and existence to an even longer causal event cone stretching back to the big bang. Making comparative statements about the efficiency of systems without including the energy needed to build them up from nothing is and always will be meaningless unless you are simply budgeting for you PG&E bill.

And this I am afraid is the real issue for AI or for the development of any complex system. How effectively does the new system use the resources at its disposal, and how effectively does the new system prepare the universe for greater complexities? What, as a matter of understanding, is "complexity"? Efficiency? Change? What makes any now different from any past or any future? Is intelligence really something that can be defined separately from the process that creates complexity?

Randall Reetz

It's an interesting demo and presentation.... However, The exchange(signal, sign, and .....) energy that define and form intelligence and life might be based on Bosanic String Theory concepts with 10-26 dimensions. That's my thinking.

Cuong Nguyen

Guong,

I have no idea what you are saying. Please take the trouble to expand your point. Use simple language. God knows this topic is complex enough. No need to further obfuscate through terse statements and non-standard use. Just write to us like we are kids. If you say something that might be confusing, use a few more sentences to explain your self. The point is communication.

Thank you,

Randall

Randall
Randall, spiritual is not a word that I would use to describe my beliefs. No, nada, nope! I don’t believe it! The definition of spirit is to week to be applicable to AI. It at best offers vague direction. This direction may offer to some something for human comfort (if we don’t have to look to deep) but it suffers from floating feet syndrome. Nothing is grounded to the degree that code can be written, that is if any of it has truth or actually applies. I do feel strongly that the direction I have taken has answers that are meaningful. My purposes of being here is either to bring them forward or mark them answered, without substance and leave them behind. I have been doing this for 35 years and it simply will propel us forward or hopefully I will understand why not. Either way it will be a tremendous relieve. I would like to ask: If on the off chance someone does produce AI gold. What happens after we are finished carrying them around on our shoulders? What next? What we look at one-another say “well I’ll be damned” and go home?
Lex

Kev
I agree regarding your statements about the tradeoff between hard-coding and learning. I might be misunderstanding what you mean by “following social norms” in this context. Things, which are social, have basic origins as well. Genetically encoding enables us make the range of sounds we are capable of. The coincidence involved when we as infants coo draws attention to us and reinforces the behavior. This provides evolutionary purpose for making sounds. However, it’s the reinforcement part that gives meaning to that sequence of events. It is from this that we find understanding and usefulness for these sounds. That would be missing from a direct programmed to do so model. This is the part of reductionism that doesn’t work. It doesn’t bring about or foster thought. It tries to superficially mimic thought.

Regarding costs, in an important sense, we aren’t there yet. It is not known what is being purchased let alone what it will cost. If you are referring to costs with regards to resource efficiency than learning will be a much more effective use of resource than standard reductionest programming techniques. In your case you would have to predefine and program it when to use, how to use, why to use, on and on. Using learning the entity will use the same basic subroutines it uses when it learns language or anything else. It will not learn anything that is not necessary for its survival. This will reduce hardware requirements. Also, it will be interesting to see if we store information ounce and refer to it when we reencounter it or are items remembered as part of events and re-stored as part of that event. This is something that we can work with but evolution might not be concerned with.

Randall
I would like to ask: If on the off chance someone does produce AI gold. What happens after we are finished carrying them around on our shoulders? What next? What we look at one-another say “well I’ll be damned” and go home?
Lex

OK Lex, this is a great topic. You ask what "we" will do after AI appears. And I think the whole notion of evolution really begs a much different question. What roll does the system that creates a new level complexity play after that new level of complexity exists?

There is something in the way your question focuses attention on "we" that triggers all of my hubris alarms. It is as though you want to talk evolution but are enable to accept its causal implications. In a system that creates novelty towards complexity handling (evolution), any entity that hope to control the future from within the limits of its own complexity handling capacity will eventually become food or fuel for another system willing to play by the rules of causality.

The honest answer: What we do (after self evolving AI exists) simply won't matter anymore! Another way of stating this acceptance: If what we do after AI exists does matter, we really haven't built or set into motion true AI.

The above answers to your question expose my own bias. I really do not care about that branch of AI that hopes to build appliances and mechanical servants. These things will be built. Market pressures alone guarantee their development. What we do after AI will be all of the things we have done before AI. We will still be the same. We will still want the same basic things. Our new smarter environment will most probably accelerate that which we imagine and this will change the scope of our imagination and of the basic materials and processes available from which to build towards the satisfaction of our desires. But all of this bounded as it is today, by the historical and causal contingencies of our own structure and resulting limits.

Of interest to me is not a THING called AI, but a profoundly novel and accelerating POTENTIAL to produce complexity. This potentiality would have to be comparative to what RNA did for the transition from freeform brownian chemistry to the meta-construction mechanics that is biology!

No AI we build as a "thing" will ever be more complex than us (see Godel). We won't be any different as a result of true AI than carbon was after RNA. On the other hand, it is impossible to extricate and isolate the evolutionary futures of two co-evolving systems. But all systems have limitations, and this is what fuels the development of new less restrictive complexities.

Randall Reet

Kev
I agree regarding your statements about the tradeoff between hard-coding and learning. I might be misunderstanding what you mean by “following social norms” in this context...

I just mean copy what everyone else is doing by “following social norms”, humans and apes are good at that (although the behavior can be suboptimal).

Regarding costs, in an important sense, we aren’t there yet. It is not known what is being purchased let alone what it will cost. If you are referring to costs with regards to resource efficiency than learning will be a much more effective use of resource than standard reductionest programming techniques. In your case you would have to predefine and program it when to use, how to use, why to use, on and on. Using learning the entity will use the same basic subroutines it uses when it learns language or anything else. It will not learn anything that is not necessary for its survival. This will reduce hardware requirements. Also, it will be interesting to see if we store information ounce and refer to it when we reencounter it or are items remembered as part of events and re-stored as part of that event. This is something that we can work with but evolution might not be concerned with.

"cost" is a multi-dimensional function, but you can boil it down to a time/energy spent vs success (not $).

If you only have to "learn" once and can deploy repeatedly you can justify a high up-front cost.

The current trends in Silicon are to produce more generic parts and program them for use in-situ (due to the high design cost [NRE]). Software loaded into the machine may or may not be self-modifiable, newer FPGAs can do partial re-programming while in use.

Being smart and learning are two different things. Humans start off knowing nothing and may cease learning once they are sufficiently expert. A machine can be smart without learning since it can acquire expertise directly from others.

Lex and Kevin,

Thanks for the thoughts. Interesting dialectic. Kevin is in a big hurry. Lex says urgency doesn't help if you don't understand the basics. Kevin shows obvious sci-fi paranoia syndrome; big bad things with terribly simplistic ideas and limitless access to data will take over everything and kill or enslave everyone. Lex places inordinate almost spiritual praise on human ability and capacity.

I'm just saying: if you use evolutionary principals in the same way nature does you are more likely to end up with an artificial person than an artificial intelligence - and their behavior can be a bit unpredictable.

Kevin, in particular you seem always to think of the present state of a system to be definitive for that system. You apply sweeping definitions to computing based entirely it would seem on the current incarnation. Computing is really only about 80 years old. I have trouble accepting that the current state of computing or even for that matter, the arch that has defined its first 80 years, in vary many ways defines the full topological or causal scope that is possible or definitive.

Actually I'm working on changing the state - http://www.parallel.c...

...snip...

And this I am afraid is the real issue for AI or for the development of any complex system. How effectively does the new system use the resources at its disposal, and how effectively does the new system prepare the universe for greater complexities? What, as a matter of understanding, is "complexity"? Efficiency? Change? What makes any now different from any past or any future? Is intelligence really something that can be defined separately from the process that creates complexity?

Randall Reetz

I would define "intelligence" as the ability to make decisions using past experience and/or future expected behavior to get better results. Better knowledge (experience) and better predicative ability makes the machine (whether biological or mechanical) more intelligent.

... I would like to ask: If on the off chance someone does produce AI gold. What happens after we are finished carrying them around on our shoulders? What next? What we look at one-another say “well I’ll be damned” and go home?
Lex

I think it may be necessary to split "intelligence" and "creativity" in this discussion. Folks can be quite bright without being very creative. I suspect that the first AIs will be bright but not creative, e.g. Monica's model-free language analysis may be able to fully understand English, but I doubt it will be writing poetry anytime soon.

I don't think we all get to go home until the AIs are also creative - the paranoid might say that we would be redundant at that point.

Kevin said: "I'm just saying: if you use evolutionary principals in the same way nature does you are more likely to end up with an artificial person than an artificial intelligence - and their behavior can be a bit unpredictable.
"

This is an interesting bias. What is it about evolution that you think includes things not human, and excludes other things done or created by us humans? This distinction is just plain wrong. If you think it isn't, please show me the line in the sand that delineates natural from un-natural, the line that that delineates the results of evolution from the the results of human thinking or human tinkering.

Here is what I know: If it looks like what a system does, any system, anything it does, is not evolutionary or "natural", then you can be absolutely sure you don't understand evolution or the causal mechanics of the system in focus.

The tendency to see humans and the product of humanity as artificial is an common mistake. The mind is so easily black-box-ed… especially when the machine used to look at the mind is the ragged edge, the nosebleed hight, of abstraction that we call consciousness.

I am further dismayed by your existential separation of the words "creativity" and "intelligence". I don't have time or the energy to go back to the basics at this moment, but every process is creative and all systems are to some extent intelligent. This discussion, any discussion about AI, is meaningless unless it is at heart a discussion about systems at least as complex as human intelligence and at least as creative. Assumed also is a level of autonomy equivalent to that enjoyed by human minds.

Maybe your interests fall towards so-called "knowledge" or "expert" systems, towards database aids as are currently bantered about as "semantic web" or "collective intelligence". If so, you must be frustrated by hard-AI as is discussed in this meetup. Maybe you think true AI is impossible or that the most effective path from here to AI is through mundane and pedestrian systems (expert, semantic, etc.). What motivates your interest in AI?

Randall Reetz

Randall, your hubris alarms may be appropriately sounding. I am proud of what I am able to understand. The sin of pride is Christian paranoia. In their eyes people with pride control others in a way Christians cannot and than do so without much wisdom. It’s clearly a Christian problem. Anyway, what I meant by “what happens afterwards” is more of an immediate concern. My question is more along the lines of what happened to the kid who pointed out that the emperor had no clothing. You may have noticed some of my own bias. My ideas are simple. The relevance of them has clearly gone under the radar. I have a penchant for the humanist model. I feel I can bring this down to the level of bits and bites. The basic idea is simplicity itself. But, you have heard claims such as this before. Some of my thoughts will make some look good, mostly others won’t fair so well. Some will look like intellectual wannabes. Others have very creative ideas that seem to coincide with my own. However, they seem to not be grounded, are too complex for immediate use and are within “the wise men describing an elephant” perspective. I’ve noticed that some of the members of this group are CEOs. Most all of us are fairly intelligent. As intelligent people, is the spirit of the group to share ideas and use them as we will? Or is this a process of pulling together to form a coalition of sorts. I don’t mind the sharing of ideas. But in the end, the question is will the robots serve us or will we once again be empowering those who require our services, robotic or otherwise.

I agree with you regarding effects of AI systems. If they only served to increase the scope of our imagination that would be enough. However, understanding thought to the degree necessary to write code that repeats its capability will allow more of the complexity you speak of. Also, I see the complexity you refer to as being a product of machine bandwidth and not a description of the functionality necessary to achieve it. e.g. An 8-bit microprocessor can do most of what a 64-bit can do , just takes longer, but within them they function almost the same. However, your next level of complexity will be from something that has very simple beginnings. If I understand you right, your assumption of intelligence is along these lines: I understand complexities; therefore, I am complex. A gadget that will be able to do the same as I, therefore, will also have to be complex. Therefore, We must look to complexities. Well, it isn’t that there aren’t complex things and ideas out there, but complex is how things end not how they begin. Complex is what the device acts upon, not an indication of its function.

I agree with what you said about the other side of AI. It will be pretty much the same. The only thing I would add is that if we are duplicating human abilities we will have to understand them first. So this has a high potential for increasing the number of self help books. Be warned!