How AI Will Revolutionize STEM and Society

[Greg Bernhardt]

We asked our PF Advisors “How do you see the rise in AI affecting STEM in the lab, classroom, industry, and or in everyday society?”. We got so many great responses we need to split them into parts, here are the first several. Enjoy!

Ranger Mike
If things run true as before, and I have seen no vast improvement in the correct forecasting of future trends from these areas, I see lots of money going in these areas but not much usable product coming out. I chose not to dwell on such predictions like back in the early 1980s when we were told the factory of the future would be a lights out manufacturing trend with only a few humans doing maintenance to keep the machines running. That and America being a service economy only in future did not prove true nor did...

Continue reading...

 

[fresh_42]

This means I won't have to answer? I didn't so far, as I have nothing profound to say about it.

 

[berkeman]

Same here. Glad the other troops stepped up!

 

[George Jones]

I did not participate, as my algorithms could not process this, and returned NaN.

 

[Borek]

Sigh, I feel the subject is so speculative ATM I preferred to not even try.

Not that I doubt we will see more and more AI in the future. I am just not sure what it will really be.

 

[Doc Al]

Same for me. Now I don't have to think of anything clever about it.

 

[phinds]

This means I won't have to answer? I didn't so far, as I have nothing profound to say about it.

Same here. Glad the other troops stepped up!

I did not participate, as my algorithms could not process this, and returned NaN.

Sigh, I feel the subject is so speculative ATM I preferred to not even try.

Not that I doubt we will see more and more AI in the future. I am just not sure what it will really be.

Same for me. Now I don't have to think of anything clever about it.

Since I have no idea what I'm talking about I just went ahead and speculated.

 

[gleem]

Wait. I don't know a great deal about the science of AI either but I do routinely follow developments as you might know because of thread I introduced last year "Notable Accomplishment in AI".

OK, AI is a bit speculative especially AGI. But the impact will still be great even if AGI is not attainable. AI can be a tool for the solution to an almost infinite number of varied problems or an instrument for the creation of some very serious ones. Considering this I am surprised that members of this forum are so disinterested.

So yeah I'm willing to stick my neck out a bit.

 

[phinds]

Considering this I am surprised that members of this forum are so disinterested.

I don't see it as lack of interest, I see it as an unwillingness to speculate by some people which is quite different. Besides, Greg said the response was amazing. How do you interpret that as "disinterested"?

 

[fresh_42]

I don't see it as lack of interest, I see it as an unwillingness to speculate by some people which is quite different.

My problem was, that I am not quite sure where algorithmic determinism and data analysis ends and AI begins. I doubt that all programs labeled AI deserve this categorization. So it wasn't the unwillingness to speculate, rather was it the unwillingness to check whether and where the label is justified. I have seen too many pigs chased through the village in my life as we say here to take any of those for serious in advance. The odds that it is just the latest trend and nothing substantial behind it, yet, are simply to high. I'd rather read a scientific paper in Computer Science than judging whether a program is AI or not.

 

[Ibix]

My problem was, that I am not quite sure where algorithmic determinism and data analysis ends and AI begins.

Perhaps we could write some kind of learning algorithm to help with that task...

 

[gleem]

Sorry Greg I think I had a mental flatulation. I was so taken aback by the posts in this thread that I overlooked (dont't ask me how) your statement.

@Greg Bernhardt I have begun a response, do I send it to you or post in the thread when it is created.

 

[Greg Bernhardt]

@Greg Bernhardt I have begun a response, do I send it to you or post in the thread when it is created.

Great! You can send it to me, thanks!

 

[vanhees71]

My problem is that I know not enough to make a statement about AI. From time to time I've heard talks about "deep learning algorithms" used as a special data-analysis technique, but that's all I've so far heard about it from non-popular sources.

 

[Astronuc]

Here's a nice summary.
https://www.the-scientist.com/magazine-issue/artificial-intelligence-versus-neural-networks-65802
"AI refers to any machine that is able to replicate human cognitive skills, such as problem solving," pattern recognition, or signal/signature analysis.

AI programs are found at many universities and scientific institutions. It has evolved with the advancement of computational resources, i.e., systems with faster and more numerous processors.

A bit more detail on neural networks.
http://news.mit.edu/2017/explained-neural-networks-deep-learning-0414

 

[fresh_42]

IMO AI should be defined as capable to lie on purpose.

 

[Astronuc]

AI is simply a tool. It can be used (for beneficial purposes) or misused (for harmful purposes).

 

[russ_watters]

Since I have no idea what I'm talking about I just went ahead and speculated.

When I think a question is too broad/vague or speculative, I just ignore it and ask and answer my own!

 

[chemisttree]

Heck, I thought everyone here but me is AI anyway...

If I prick you, do you not... short out?

 

[phinds]

Heck, I thought everyone here but me is AI anyway...

If I prick you, do you not... short out?

I don't know about the other guys here but if you try to prick me, I'm calling the cops !

 

[f95toli]

My problem was, that I am not quite sure where algorithmic determinism and data analysis ends and AI begins. I doubt that all programs labeled AI deserve this categorization. So it wasn't the unwillingness to speculate, rather was it the unwillingness to check whether and where the label is justified. I have seen too many pigs chased through the village in my life as we say here to take any of those for serious in advance. The odds that it is just the latest trend and nothing substantial behind it, yet, are simply to high. I'd rather read a scientific paper in Computer Science than judging whether a program is AI or not.

I agree with. In my field (experimental SC quantum computing) AI comes up in two contextsL

1) In helping run experiments. Primarily to optimise experimental parameters "on the fly" when dealing with very large parameter spaces (which is very often the case) and/or in feedback. Here the "boundaries" between AI, ML and regular optimisation are very blurry.
See e.g.
Lennon, D.T., Moon, H., Camenzind, L.C. et al. Efficiently measuring a quantum device using machine learning. npj Quantum Inf 5, 79 (2019). https://doi.org/10.1038/s41534-019-0193-4

However, some of the "ML" work is little more than clever curve fitting, so I wouldn't count that as AI.

2) As a possible application. Quantum machine learning is quite a hot topic at the moment (e.g. image recognition) and there is also work on things like deep learning (which I guess is more like "proper" AI)

 

[StatGuy2000]

My problem was, that I am not quite sure where algorithmic determinism and data analysis ends and AI begins. I doubt that all programs labeled AI deserve this categorization. So it wasn't the unwillingness to speculate, rather was it the unwillingness to check whether and where the label is justified. I have seen too many pigs chased through the village in my life as we say here to take any of those for serious in advance. The odds that it is just the latest trend and nothing substantial behind it, yet, are simply to high. I'd rather read a scientific paper in Computer Science than judging whether a program is AI or not.

You raise a good point, which is connected to my own reply in the Conversations regarding AI.

I find that discussions regarding AI are often confused, because there are a number of different definitions on what artificial intelligence actually encompass.

According to Chapter 1 of the best-selling and standard textbook on AI -- Artificial Intelligence: A Modern Approach, by UC Berkeley computer scientist Stuart Russell and Google researcher Peter Norvig (a copy of which I own) -- AI can be defined in four different areas:

1. Thinking humanly
2. Thinking rationally
3. Acting humanly
4. Acting rationally

#1 and #2 together are concerned with thought processes and reasoning, whereas #3 and #4 together are concerned with behaviour.

#1 and #3 together measure success in terms of fidelity to human performance, whereas #2 and #4 together measure against an ideal performance measure, which we call rationality.

Any discussions regarding what AI is or is not needs to take the above definitions into account.

 

[gleem]

"A rose by any other name would smell as sweet" It doesn't matter what you call it, machine learning, expert systems, natural language processing, intelligent retrieval, artificial learning, etc. Let us not get hung up on what we call AI and just concentrate on why, how, and where this technology can or should be implemented. Concentrate on the benefits and risks of its implementation. Develop plans and strategies to maximize the benefits relative to risks before things get out of hand and we end up with untenable problems.

 

[phinds]

IMO AI should be defined as capable to lie on purpose.

I think that's an end game scenario. We'll have highly versatile, useful, and productive AI LONG before we get to that.

 

[fresh_42]

I think that's an end game scenario. We'll have highly versatile, useful, and productive AI LONG before we get to that.

I'm still insecure whether AI isn't simply a better data analysis method rather than Intelligence. It isn't that long ago that we refused the animal world to have any intelligence beyond instinct and heritage at all! At the same time we started to use the term AI? This doesn't make much sense to me. Some species are capable to lie. Yet, we refused to call it Intelligence. And now some data mining tools should be called as such? Ridiculous.

 

[phinds]

I'm still insecure whether AI isn't simply a better data analysis method rather than Intelligence. It isn't that long ago that we refused the animal world to have any intelligence beyond instinct and heritage at all! At the same time we started to use the term AI? This doesn't make much sense to me. Some species are capable to lie. Yet, we refused to call it Intelligence. And now some data mining tools should be called as such? Ridiculous.

I think it gets down to just arguing about terminology. I'm an engineer. I'm more interested in actual characteristics than I am in names.

 

[fresh_42]

I think it gets down to just arguing about terminology. I'm an engineer. I'm more interested in actual characteristics than I am in names.

Not really. The question is: What makes AI different from all classical algorithms? Is it a new property or a only a sophisticated algorithm?

But I admit that I'm playing the devil's advocate in this discussion and not all of my statements would survive a further analysis. Personally, I don't like the "I" in AI. To me this is a sign of the usual human arrogance rather than a justified name. And a point of sales. So far we have had (probably among others):
ERP's (80's), IT certification (90's), OO (90's), Customer Relationship Management (90's), Risk Analysis and Backup Systems (00's), Data Mining (00's), Cloud Computing and other remote concepts (10's) and now it is quantum computing and AI. It keeps consultancies busy and fills team meetings.

In this sense it is about terminology. But is also the question whether the word Intelligence is justified, especially under the scope that we refuse(d) to label Intelligence in non human nature. If terminology suggest a property which isn't there, then terminology becomes an issue.

 

[gleem]

@fresh_42 You may be interested in one of the first AI projects (programs) call the "Logic Theorist" developed by A. Newell, J C Shaw, and H A Simon. Newell was a cognitive psychologist. Among other things, they used it to prove theorems from Ch 2 of Whitehead and Russell's "Mathematica Principia".

You can download it at https://www.rand.org/pubs/papers/P1320.html

 

[jack action]

The question is: What makes AI different from all classical algorithms? Is it a new property or a only a sophisticated algorithm?

No matter what is the answer to those questions, The question that was asked was:«How do you see the rise in A.I. affecting STEM in the lab, classroom, industry and or in everyday society?»

One point I raised in my answer is the fact that a lot of people want to use AI (or whatever you want to call that technology) to make final decisions, without any human supervision (like driving a car or maybe even make societal choices). If we do go down that road (should we allow it?), we will have to redefine what responsibilities one individual has towards another one or even the group.

You must have an opinion about that.

 

[fresh_42]

You must have an opinion about that.

I do. I would consider it as an additional parameter rather than a trigger for decisions. But I'm a bit old fashioned and things are far less dramatic in reality than Asimov's laws would suggest. It is not the apocalypse waiting around the corner, neither the futurological congress, nor skynet, nor the matrix. It will be a useful tool and certainly not all fantasies will come true. I am strongly against decision making by machines, It causes people to drive into rivers because GPS tells them to do. Just a few hours ago, Cortana informed me where she (?) put my alerts into. As if that was an "I". If I knew how, I would deinstall that rubbish.