Category: Math

The Past, Present, and Futures of Artificial Intelligence

Tik-Tok — one of the early visions of Artificial Intelligence

Artificial Intelligence is in the news, no question. The last few science fiction conventions I’ve been at, panels on AI have been getting filled rooms & lots of questions. It’s all over the news as well. And it’s a subject I find interesting: and — being a professional programmer — an area where I may be able to contribute, perhaps looking at the use of AI techniques to generate physics experiments.

What is meant by AI is one problem: is it anything that uses AI techniques, as Neural Nets or Genetic Algorithms? Or do you need to be pointing in the direction of some kind of sentience for it to be true AI? Will it replace us? Should it? The hype/content ratio sometimes hits near Trumpian levels.

So good subject for a talk. But what line of attack to take? Last week I caught a great talk at DataPhilly on the use of AI for sports betting (and other things). The formal title was:

Practical Scaling: How to Use Simple Tools to Create and Implement Complex Modeling SystemsJames Piette

For “complex modeling systems” think practical AI. My personal favorite of his slides cited three principles:

  • Moravec’s Paradox: AI and humans are good at opposite things. So use AI for what it is good at (crunching) and let the humans do what they are good at (intuiting).
  • Pareto’s Principle or the 80/20 rule. 20% of the work gets you 80% of the value, so focus on the high value parts of the problem.
  • The Scientific method: observe, hypothesize, test, repeat. This works for science — and it works for software debugging (another favorite topic of mine), startups, and AI.

And this strikes me as the kind of no-nonsense, practical, even scientific approach that a subject like AI needs. (Thanks James!) So for my 2020 SF Convention talk:

The Past, Present, and Futures of Artificial Intelligence. — From Oz’s Tik-Tok to the Mechanical Turk, from Neural Nets & Genetic Algorithms to Chess & StarCraft, from Medical Diagnosis to Robot Frogs, from Facial Recognition to Fakes, Deep Fakes, & Anti-Fakes, AI is everywhere today. How did it start? What do we mean by AI? What are the basic AI techniques? How is it being used? What are the benefits? risks? and how should we manage AI going forwards?

Be seeing you:

Is time fuzzy?

Alice’s Past is Bob’s Future. And vice versa. Both are bit fuzzy about time.

“Time dispersion and quantum mechanics”, my long paper — long in page count & long in time taken to come to completion — has just been accepted for publication in the peer-reviewed Proceedings of the IARD 2018. This will be has been published as part of the IOP Science’s Journal of Physics Conference Series.

I had earlier presented this as a talk at the IARD 2018 conference in June 2018 in Yucatan. The IARD (International Association for Relativistic Dynamics) asked the conference participants if they would submit papers (based on the talks) for the conference proceedings. No problem; the talk was itself based on a paper I had just finished. Of course the paper had more math. Much much more math (well north of 500 equations if you insist).

Close review of the talk revealed one or two soft spots; fixing them consumed more time than I had hoped. But I submitted — on the last possible day, November 30th, 2018. After a month and a bit, the two reviewers got back to me: liked the ideas, deplored the lack of sufficient connection to the literature, and in the case of Reviewer #1, felt that there were various points of ambiguity and omission which needed attention.

And right they were! I spent a few rather pleasant weeks diving into the literature; some I had read before, some frankly I had not given the attention that must be paid. I clarified, literated, disambiguated, and simplified over the next six or seven weeks, submitting a much revised version on Mar 11th this year. Nearly ten per cent shorter. No soft spots. Still a lot of equations (but just south of 500 this time). Every single one checked, rechecked, & cross-checked. And a few fun bits, just to keep things not too dry. Submitted feeling sure that I had done my best but not sure if that was best enough.

And I have just this morning received the very welcome news it will be joining the flock of accepted submissions headed for inclusion in the conference proceedings. I am best pleased.

As to the title of this blog post, my very long paper argues that if we apply quantum mechanics along the time dimension — and Einstein & even Bohr say we should! — then everything should be just a little bit fuzzy in time. But if you title a paper “Is time fuzzy?”, you can say farewell to any chance of acceptance by a serious publication.

But the point is not that time might be fuzzy — we have all suspected something of the kind — it is that this idea can be worked out in detail, in a self-consistent way, in a way that is consistent with all experimental evidence to date, in a way that can be tested itself, and in a way that is definitive: if the experiments proposed don’t show that time is fuzzy, then time is not fuzzy. (As Yoda likes to say: fuzz or no fuzz, there is no “just a little-bit-fuzzy if you please”!)

In any case, if you are going to be down Baltimore way come this coming Memorial Day weekend I will be doing a popular version of the paper at the 2019 Baltimore Science Fiction convention: no equations (well almost no equations), some animations, and I hope a bit of fun with time!

The link at the start of this post points to a version formatted for US Letter, with table of contents & page numbers. The version accepted is the same, but formatted for A4 and without the TOC and page numbers (that being how the IOP likes its papers formatted). For those who prefer A4:


Relativistic Morlet Wavelets

Wavelets are like musical notes: they are wave forms limited in both time and frequency. What makes them particularly useful is that any reasonable wave function may be written as a sum over them.

Usually we think of music in terms of pure tones, in terms of its Fourier components. But pure tones can be a bit too pure. For one thing, if a tone is to be completely pure it has to last forever, not a characteristic associated with practical questions. Wavelets are impure tones, and therefore a better match to the real world.
Read more »

WordPress Themes