AI and Machine Learning

[sengsational]

I don't know what to think about computers getting "smarter"; If you had told me 25 years ago what's happening now, I would have doubted you. The exponential advancement of technology will mean that I'll probably doubt even more a prediction you tell me now, even though I know I shouldn't apply linear thinking.

The reason I brought this up is a page on NYT. If you're a subscriber or haven't run out of the bits they give you for free, this is worth a look, IMO.

It shows a picture of a person morphing from one person to another. We've all seen that before, but these people don't exist. They're created by a machine. There's also some sliders to morph gender, mood, etc. And it ends with a way to spot the fakes.

https://www.nytimes.com/interactive...rtificial-intelligence-fake-people-faces.html

 

[Out-to-Lunch]

I don't know what to think about computers getting "smarter"; If you had told me 25 years ago what's happening now, I would have doubted you. The exponential advancement of technology will mean that I'll probably doubt even more a prediction you tell me now, even though I know I shouldn't apply linear thinking.

The reason I brought this up is a page on NYT. If you're a subscriber or haven't run out of the bits they give you for free, this is worth a look, IMO.

It shows a picture of a person morphing from one person to another. We've all seen that before, but these people don't exist. They're created by a machine. There's also some sliders to morph gender, mood, etc. And it ends with a way to spot the fakes.

https://www.nytimes.com/interactive...rtificial-intelligence-fake-people-faces.html

Wow, that was pretty amazing.

Was it wrong of me that I lingered on the particularly attractive ones so long?

 

[sengsational]

Was it wrong of me that I lingered on the particularly attractive ones so long?

The animal brain. Embrace it.

 

[Lsbcal]

In general predictions are interesting but not to be taken too literally. So predicting where all this is headed is fun but not very likely to be right. I constantly remind myself about the limits of models going forwards in time. Weather forecasting and stock modeling are good examples of limited models. I would be quite rich if I could predict stock prices a second or so in advance.

In physics we know that we cannot perfectly solve problems with even a few idealized bodies present. For example, this article https://en.wikipedia.org/wiki/Three-body_problem

And that is if we know the starting conditions which in the real world are only known to limited precision.

One solution to computers getting smarter is for us to become better users of the information. Can we do that as individuals? I think some here can but I don't know about our society in general.

 

[yakers]

Just got off a Zoom briefing from the 'futurologist' at The Economist Magazine. Last year just for an article idea he interviewed an AI futurologist, turns out 3 out of 3 predictions by the AI came to pass, much better than his record.

 

[Sunset]

Wired magazine has an interesting use of AI , by our soldiers to clear buildings in dangerous places.

 

[teetee]

The recent news about AI is the invention of AlphaFold a protein folding prediction algorithm that is able to apply large range of proteins without using the amino acid sequence homology inputs which almost all current tools use (Folding@home) with a good results comparing to using crystalization method (40yr+ old wet bench method to learn about the structure of protein and other macro molecules).

Similar to the fake face creation, AlphaFold is just one of the applications that use deep machine learning to train itself to make better decisions.

 

[Pellice]

I know there's a ChatGPT thread that is very active and interesting, but this is the only thread I could (quickly) find that is about the somewhat broader category of artificial intelligence (mods, feel free to merge if you think best). What's interesting to me is that this thread, begun by Sengsational in late 2020, had only 7(!) responses.

How times have changed.

Do we have a better idea now, a little over 2 years later, as to how this new technology will develop? The Hill[/I ]online website has an opinion piece taking a dire view: "Artificial intelligence will destroy ‘laptop class’ workers"

The thesis of this piece:

Scores of jobs that require a college education will be changed nearly overnight. Rapid advances in this new technology will wreak havoc on the very people who prospered during COVID, especially those who work in the “knowledge economy” and can often carry out their duties from their laptops at home.

I think this is overblown. I am not convinced. But certainly the technology is warping ahead right now.