Proposing FTL today is pretty much a lost cause (although there are interesting ideas such as the Alcubierre Drive – at least it’s still a viable option for Sci-Fi literature). The problem is that the Einstein’s special theory of relativity (SRT) postulates that nothing is able to go faster than light (more exactly: no information can be transmitted faster than light. Spaceships and their passengers are of course matter organized in certain ways, viz information – so they can’t go faster than light).

To date, we have no observations which contradict relativity, on the contrary, SRT and GRT are highly successful and thoroughly corroborated. To put it another way, observation of an FTL object would be theoretically quite…a surprise.

The situation is very different for strong AI: first of all, there is no theory whatever which predicts that such a thing were not possible. We don’t know quite exactly how this thing called consciousness appears in the brain, but it is a very active subject of research.

But there is something more important: whereas we have not observed FTL objects, we observe strong “I’s” (intelligences) every day: your fellow humans, yourself etc.

We are conscious, and we live in this physical world, made of the same physical stuff as everything else. Our consciousness is an organizational property of the matter we are made of, not something magical tacked on as an afterthought. What a wonderful insight: we know by simple observation of our surroundings that physical matter configurations can become conscious! Easy to see, yes? But, as Aristoteles said: “Just as bats’ eyes are to daylight, so is the mind blind to that which is most obvious of all.”

Strong AI is not a problem in the sense of “could it possibly exist?”; it is evidently only an engineering problem (albeit a complex one). Maybe we will need molecular biology for solving it (meaning that AI will only run on proteins and not on silicon, which is more in line with materialism than computationalism; but it is engineering nonetheless). We just have to find out in which way matter has to interact (in a sufficiently reentrant way) to build an AI.

After all, the distinction artificial and natural is pretty thin anyway. Ants are natural. Their nests are natural. Humans are evolved, they are natural. So why should we not call their artifacts natural? It is only a philosophical word quibble – the distinction natural/artifact is sometimes interesting – for instance, when we stumble upon an artifact on an alien world, this is astounding not because an “artifact” were something “supernatural”, totally out of this world, but because an “artifact” suggests an “artificer” -  an intelligence, an agent, which made it – a natural being of some sophistication. The intelligence that made the thing would be quite a natural inhabitant of its environment. Never let the distinction artificial/natural confuse you!

So, to get back on topic: what opponents of strong AI would actually have to claim is that we will never (1000 years? 1 million years? our descendants on different planets in 5 billion years?) be able to engineer a conscious artifact because of mysterious reasons which would go against everything we know about this world. And this claim, now, seems completely ludicrous to me. And if it does not sound ludicrous to you, go look in a mirror! (you are conscious, are made of matter, but are not going faster than light ) Opponents of strong AI are fighting the same losing battle as vitalists did in the 19th/20th century.

I found this amusing quote by Francis Crick on the wikipedia article on vitalism: “And so to those of you who may be vitalists I would make this prophecy: what everyone believed yesterday, and you believe today, only cranks will believe tomorrow.”

Similar issues are raised in this blog post, which motivated me to publish this (now slightly overworked) draft (which otherwise would have slumbered for many months on my harddrive before being polished enough to publish ).

A commenter on that blog (Accelerating Future) says this:

Michael Bishop (2003). Dancing with pixies: Strong artificial intelligence and panpsychism

This paper argues against computationalism by showing it implies panpsychism.

Here it should be emphasized that in fact every realist monist naturalism (and who’s a dualist nowadays? nobody, for good reasons!) implies panpsychism – see this paper by Galen Strawson. (If you want more of this stuff, there’s also a book.) So the problem (if it is one) lies somewhere else and certainly not in positing strong AI or computationalism. The problem can be solved by a radical (“radical” in the sense of “going to the root”) monism, but on that more later, because it’s a topic of its own.

Technorati Tags: consciousness, human brain, materialism, philosophy, strong ai

Overcoming Bias: Heading Toward Morality

Why? Because that is the conclusion he is drawing to:

Artificial Intelligence melts people’s brains. Metamorality melts people’s brains. Trying to think about AI and metamorality at the same time can cause people’s brains to spontaneously combust and burn for years, emitting toxic smoke – don’t laugh, I’ve seen it happen multiple times.

But the discipline imposed by Artificial Intelligence is this: you cannot escape into things that are “self-evident” or “obvious”. That doesn’t stop people from trying, but the programs don’t work. Every thought has to be computed somehow, by transistors made of mere quarks, and not by moral self-evidence to some ghost in the machine.

If what you care about is rescuing toddlers from burning orphanages, I don’t think you will find many moral surprises here; my metamorality adds up to moral normality, as it should.

snip

Yet there is also a good deal of needless despair and misguided fear of science, stemming from notions such as, “Science tells us the universe is empty of morality”. This is damage done by a confused metamorality that fails to add up to moral normality. For that I hope to write down a counterspell of understanding. Existential depression has always annoyed me; it is one of the world’s most pointless forms of suffering.

The last paragraph nicely illustrates the goal of my thesis. Much of resistance encountered to scientific thinking stems from confused thinking about existential questions, morality and the meaning of life etc. This pre-scientific – mostly magical and inconsistent – thinking can and should be dispelled.

On a side note:

As you will have noticed I often quote Eli’s posts – simply because he has eloquently written down what I only have been thinking about. Most of the times when I want to write something down I realize I can’t explain it in a few sentences or even a few paragraphs – NOT because it is confused or not coherent yet, but rather because after years of thinking on these subjects I have dissolved many mental categories, rewired others and created new ones. Then I fall into a kind of despair and write nothing (I really have to work on this).

The funny thing is that Eli Yudkowsky (and many others who also have not written yet, but are among the OB readers or elsewhere strewn over the internet) have undergone similar mental evolutions. So much for multiple discovery.

While many scientists and philosophers have “problems” with multiple discovery because they are concerned over priority, I think it is a happy thought that ideas are discovered simultaneously in many places when the time is ripe. It shows that intellectual progress is robust against individual chance events.

My main problem is how I will deal with this inferential gap issue in my thesis: while people working within traditional philosophical schools simply refer to their “founding fathers” – for instance Kant – and their advisors immediately know what they are talking about (indeed, they taught it to their students!), I can’t simply refer to Eli’s posts in my thesis – most people do not know him – not yet, anyway , maybe in 20 years time.

So I will either have to bite the bullet and write a huge expository chapter; or wait until Eli writes his book and ask my advisors to read it before I have my defensio (I guess that would be a quite cheeky approach – I’ll probably settle with the expository thing )

A last remark: why this large deviation from standard philosophy? Because standard philosophy misses some major insights, namely those gleaned from AI research of the last fifty years. Every philosopher should plow through this book (Russell & Norvig: Artificial Intelligence. A Modern Approach. 2nd Ed. 2002) before commenting on issues such as rationality or free will.

Funnily enough, Eastern Philosophy (Taoism, Buddhism) does not have this problem: that is because they stressed meditation, which is empirical/rational introspection versus the rather naive introspection performed in the West. Thus, the sages of Zen discovered many things about themselves which escaped the likes of Descartes, Kant etc.

Technorati Tags: artificial intelligence, phd, philosophy

Creativity will come from working passionately in the field – novel approaches will “pop” into your mind (actually, it’s just your brain rewiring with time )

As to IQ – I think, g factor notwithstanding, that we should be very careful in this regard. I liked Peter’s classification as IQ measuring a shared passion for solving (certain kinds of?) puzzles.

Genius, Sustained Effort, and Passion (Apperceptual)

Technorati Tags: artificial intelligence

(The essay was brought to my notice via overcomingbias.com)

Some quotes:

This sentence is gold:

The truth is common property. You can’t distinguish your group by doing things that are rational, and believing things that are true. If you want to set yourself apart from other people, you have to do things that are arbitrary, and believe things that are false.

And here:

We arrive at adulthood with a kind of truth debt. We were told a lot of lies to get us (and our parents) through our childhood. Some may have been necessary. Some probably weren’t. But we all arrive at adulthood with heads full of lies.

There’s never a point where the adults sit you down and explain all the lies they told you. They’ve forgotten most of them. So if you’re going to clear these lies out of your head, you’re going to have to do it yourself.

Few do. Most people go through life with bits of packing material adhering to their minds and never know it. You probably never can completely undo the effects of lies you were told as a kid, but it’s worth trying. I’ve found that whenever I’ve been able to undo a lie I was told, a lot of other things fell into place.

I have to absolutely second the last sentence – the most gratifiying experience when adopting a scientific mindset – or, more strongly, adopting a scientific identity – is the freedom to systematically discard the fantasies underlying your local group (nationality, ethnicity, religion, what have you…) world view: and from that point on the world starts to make more sense every day!

It’s not enough to consider your mind a blank slate. You have to consciously erase it.

It reminds me of something which I said in a lecture half a year ago: before starting to learn things, you have to unlearn most of that which you think you know (because it is false). The “knowledge” base existing in our brains is probably the biggest barrier to erkenntnis (insight, knowledge, truth) we face.

Happy cognitive deleting

Technorati Tags: artificial intelligence, bayes, bias, cognitive science, critical thinking, ethics, philosophy, rationality

The blog entry below has a few links to the central concepts of Gärdenfors’ theory. I agree with the need for integrating different levels (symbolic, connectionist and spatial in Gärdenfors’ version), but I think that ultimately one will need even more levels of modelling to capture thought processes.

Thought, I think, is the result of a veritable holarchy (thanks to Glen E. Ropella for alerting me to this concept in a mailing list discussion) of simultaneous processes in the brain with feed-forward and feed-back cycles. The holarchy would be made up of neuronal aggregates forming and breaking up spontaneously (within but also crossing modular boundaries).

Three Levels of Thought

The holarchy concept merits a post of its own – sometime

Technorati Tags: artificial intelligence

Two AI Pioneers. Two Bizarre Suicides. What Really Happened?

Technorati Tags: people

Cyborg moths Times Online

Technorati Tags: artificial intelligence, singularity

Numenta was formed to develop and promote a technology called Hierarchical Temporal Memory, or HTM. We are confident that the principles underlying HTM are the same principles that govern much of the operation of the human neocortex; thus, HTM enables the creation of machines that have some of the capabilities of the human brain. We see HTM as a fundamental new computing methodology able to solve longstanding problems in artificial intelligence and machine learning. Be sure to read the white paper on our web site that describes the basics of HTM and what it can do.

I haven’t yet found time to read the white paper, but will correct this soon

Release Notes

White Paper

Technorati Tags: artificial intelligence

Why is this important? Because their brain architecture is different from ours (humans), it may help extract necessary and sufficient conditions for intelligence and consciousness. This will hopefully further our quest toward developing a conscious AI.

Technorati Tags: artificial intelligence, cognitive science

To make one thing clear: I do not think that artificial intelligence is impossible, on the contrary, I do think that we will be able to develop an artificial intelligence. But paramount to this effort is to get it right, and a lot of AI research is dominated by wrong metaphors (ie software/hardware distinction etc).

Only when we get a clear understanding of how brains work will we be able to develop an artificial brian. Up to now, it was the other way round: we built computers and transferred the mechanisms of computing metaphorically to human brains, ie: memory = hard disk, attention span = ram, thinking = computation, brain = hardware, thoughts = software.

Technorati Tags: artificial intelligence, cognitive science