code and the oracular


leave a comment »

Dear Dad,

I’m really proud of that paper. You can hear bits of my wording creeping in, due to suggestions given to Arka when he finalised it.

This does represent a start point for using deep learning, which has been crucially productive in many different applications already, and is just beginning.

the next step which will happen (inevitably as things dependent on distributed efforts do) is to get a deep learning system that munches up movies and starts with a reward based predictive structure to guess the next line in the dialogue. This is a clear logical extension of tried and tested facial recognition concepts, where you are expanding the image recognition to take in moving, not still, images and sucking patterns and connection out of them.

There are already large dialogue corpuses available, from many sources including both tech support chat transcripts and also even movie scripts. If socially capable robots are to have a future then deep learning has to win at modelling human social interaction in the kind of framework that I came up with here. I went through a huge phase of skepticism, and then an almost Luddite antipathy to the idea. ( see ) But now I’ve reread the paper I have to say I think this technology will emerge eventually… and change the world (yet) again…

some of the ideas I used are embodied in these cartoon slides I made also

The concept is of a more refined interface for computers than we currently have which anticipates your needs better. It’s not your kind of film, but the excellent Iron Man series really shows this in the form of “Jarvis” Tony Stark’s personal assistant software.

Of course you can get hung up on the “Zombie” question posed in the paper, because no matter how you work the thoughts “Artificial Consciousness” seems an impossibility. It does to me also, I am not drinking the Kool Aid of silly dreamers who can’t see that autonomous robots are about the most difficult computing challenge we have seen. But as with people who doubted the feasibility of aeroplanes, something is bound to come of all this.

Scientific knowledge workers hunt for, and often can’t resist, challenges. This is a partial explanation also for the times science may seem to have “gone wrong”: The Manhattan Project, biological weapons, DDT, Nazi experiments… etc The failing shown here was to be so hypnotised by one’s power and the magnetic pull of the challenge itself and the joy of dreaming up functional solutions that the ethical questions were neglected. I became very anti-ai after writing this paper, but now my only idea is to accept the inevitability of increasing technological development while preserving hope that Terminator scenarios don’t come true, as a kind of Kantian investment in the principle that if I govern myself morally then it is likely others will too.

One of the things that came to me, partly caused by a general suspicion of the modern world which had grown in me after seeing how many troubles we have caused ourselves as a species, was the idea that somehow human intimacy, love, friendship etc might be cheapened by the idea of artificial friends. Technology in the modern conception starts with economic exploitation of the new ideas it brings. I saw a world of accelerating gaps between rich and poor and a new “corporatocracy” that holds huge power. Of course dystopian science fiction features that theme a lot. It started perhaps with Wells and then forged ahead with 1984 by Orwell. Since then about a third of all new movies seem to embody some kind of “disaster sci-fi”.

But one can’t sniff at science fiction any more, when one sees that it is a direct consequence of a relatively new force in human affairs: an accelerating rate of technological change. For millenia a relatively slow pace of change allowed continuity of belief between parents and their offspring. But now this force that has given rise not only to wider generational differences but wider gulfs between the enjoyers of development and those deprived too. And yet somehow some of these movies that are placed in the popular, rather than the highbrow, category are crucial. They form a backdrop of culture that colours in so many unknowns and it is no coincidence that all the authors of this paper are science fiction freaks motivated by crazy dreams that stir our imaginations so.

Again this is connected to the moral element, because just as an engineer often cannot resist a problem awaiting solution, so also a speculator cannot often resist the opportunity of profit. We need to govern our lives using every resource given us by the miracle of nature, not by becoming unbalanced in the emphasis of one faculty over another. This holism is actually what made me dream up the idea of HEAD because my dream for the systems behind head are more like a liberal democracy than a totalitarian state. Each attribute of one’s cognition is another ally, a tinkerbell speaking wise advice just at the right time, a healthy parliament where many heads are better than one… Or Minsky’s “Society of Mind”.

So also I believe that a conversational interface is a grey area between a friend and an app. If I can “speak” to a system like this in the usual channel that I often speak to my friends, say Skype, WhatsApp, SMS etc then a complex computer program has moved part of the way into a “social” zone where other humans are also present.

But to return to the theme, human contact and intimacy must not be cheapened. It is a starting condition for the infinite meanings that life offers, as it is also our first experience of reality as dependent infants, upon which our entire personhood is based. But the capability of computation to enhance our human reality is also immense. If it is truly as many have seen a Pandora’s box… or even Prometheus’ theft of fire from the gods we need to be very circumspect.

I hope we can find a balance. This seems to be one of the crucial issues of our age.


Written by Luke Dunn

September 10, 2017 at 11:17 am

Posted in Geek Teachings, robotics, transhumanism

Tagged with

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: