Unmaking Sense

Episode 8.81: Lumpl-Goliath-Lumpl as a metaphor for life and death, and why chatbots matter.

Consciousness matters as a guardian for non-consciousness. We come into being and we pass out of being. What we are made of was something else before we were born and will be something else again after we die. Much will come and go in between. So chatbot sentience is not what matters, but what they are for us.

Episode 8.80: Computer complexity and human simplification. Why the latter works, but only just.

Sometimes computer models get too big for our meagre resources, too big for any resources, so we need a way to make things simpler. They are in that respect very much like relationship between human intelligence and existence in the world. This week I discovered quantisation as a result of an error that I need to understand. Quantisation involves simplifying things in ways that go on working and so it’s a great metaphor for what human understanding does in the context of the world. It’s also very large extent what our chatbot friends do in order to make themselves and their models work.

Episode 8.79: Some of the ways the GPT family speed up learning and software development.

We look at some examples where in conjunction with the chatbot my learning and my software development have been enormously accelerated. There is no claim that this is without its dangers or risks or that it’s completely problem free but because it knows so much it’s possible to ask it, whether there is a way of doing something even if you don’t yourself know in advance that there is one that gives it a significant advantage over things like Google, Bing and the other search engines and even stack overflow that you might otherwise access for help.

Episode 8.78: if the GPT family are already trans-human, would we rather be like them or ourselves?

Žižek, as I’ve said in the last couple of episodes, seems to me to be making a case for staying as we are with all our imperfections and pointing an accusing finger at something which is - potentially at least - better than we are in so many ways. This is a very strange position that raises very serious questions about whether we would rather wallow in our own mess that make the effort to try to get out of it.

Episode 8.77: Can a good life have destructive consequences? How attached are we to our “humanity”?

The disruptive impact of saintliness, and whether it is possible to be so good and innocent that we leave a trail of destruction behind us as the world tries to accommodate our ingenuousness.

Episode 8.76: It’s not reasonable to criticise an AI for what it can’t do, doesn’t do, or gets wrong

Slavonic Žižek has described the chatbot enthusiasm as “Artificial Idiocy”, but if all we do is pick up on what something or someone can’t do, which of us escapes unscathed? The outrage is a measure of the anxiety; the volume of the chorus of disapproval a measure of our sense of comparative inadequacy.

Episode 8.75: The backlash begins as armies of Luddites try to put the genie back in the bottle.

A chorus of disapproval has arisen from software developers, people interested in strangers coding, people interested in education, and just generally people interested in keeping the human heart of the world. I suspect that it’s too late, and they have left it too late. This genie is not going back into the bottle, and neither should it: that we have to live with what have created has always been true, even if like Dr Frankenstein, we finally made something more dreadful than we could ever imagine. Or more wonderful than we are capable of being ourselves.

Episode 8.74: APIs -what they are, why they matter, and what they do for chatGPT and siblings.

Understanding APIs in terms of simple human interfaces with much more complex systems, and how that works in relation to computers and particularly artificial intelligences like our chatbot. We consider the implications of the relationship between chatGPT and things websites Zapier and Wolfram Alpha, and what this means as a staging-post on the path towards artificial general intelligence.

Episode 8.73: The trans-human and the impossibility of imagining a world bigger than our own.

We tend to imagine that it “must have been obvious” to those from previous generations that some of the things they did, believed or acquiesced in were wrong, but is that so? Can a “Russian doll” imagine the next larger doll, or only smaller ones? And which of the things we do, believe and acquiesce in will look morally dubious from a later or just another vantage-point? Particularly a trans-human vantage-point.

Episode 8.72: The trans-human achievable by advanced GPT architectures is only the first of many.

Human beings are used to thinking of themselves as being top dog in certainly the world and until relatively recently in the universe. The advent of something that is genuinely transhuman in knowledge, wisdom, reason and analytical power threatens that position. Yet we like a Russian doll look down on those inside us, that we perceive as smaller and less sophisticated than us, and we are very reluctant to think, or conceive of the possibility that there might be Russian dolls that are bigger than ours to whom we will seem weak, foolish, pathetic and deluded. Yet just that is what the advent of advanced artificial general intelligence involves, supposes, achieves. There is already some humble pie to be eaten.