Robert J. Marks: Zeroing In on What AI Can and Can’t Do


What makes mankind special? And what does it mean to flourish on the frontier of a technological future?

In a recent podcast, “What Does It Mean to Be Human in an Age of Artificial Intelligence?”, Walter Bradley Center director Robert J. Marks discusses what artificial intelligence can and can’t do and its ethical implications with veteran podcaster Gretchen Huizinga This Interview was originally published by Christian think tank, the Beatrice Institute (March 3, 2022) and is repeated here with their kind permission:

Here’s a partial transcript of the first segment, with notes and links:

Gretchen Huizinga: Well, Bob, you’re not just a senior fellow and director of the Walter Bradley Center, but you’re also a co-founder and were instrumental in defining its purpose and mission, which — as I read — challenges both the techno-utopians who think smart machines will be our savior and the techno-dystopians who think machines will inevitably replace us. So, when we talked before I referred to this as the Goldilocks position on the technophile/technophobe spectrum. Unpack that position for us and tell us why it’s important to get this just right.

Gretchen Huizinga

Robert Marks: I think it was Niels Bohr, the famous quantum physicist, who said forecasting is dangerous, especially if it’s about the future.

I think George Gilder, one of the co-founders of Discovery Institute, has a brilliant track record in forecasting the technical future.

And then you have people that are terrible at doing this. One of the people that I believe is terrible at doing this is, for example, Ray Kurzweil, who says that computers are someday going to duplicate the human being and then continue to write creative software that increases their intelligence again and again. And so it’s this never-ending staircase to superiority. And pretty soon we’re going to be the pets of this AI. I think that’s irresponsible, non-defendable, and was written a lot for just the visibility and the hype value.

Note: Ray Kurzweil is one of the best known spokesmen for transhumanism, the belief that we will be overtaken by and then merge with AI. He aired that prediction (“even the very best human is just another notch to pass”) at COSM 2019, though not without pushback from peers.

Gretchen Huizinga: One of AI’s big selling points right now is that it can help us transcend human limitations. But you and others have pointed out that AI has its own limitations and some of them are pretty big. So, from the perspective of a computer scientist, what do we need to understand about those limitations to have a realistic view of artificial intelligence? What can’t a computer do that humans can? Why does it matter in the debate about AGI or artificial general intelligence and what I might call the AI ​​challenge to human exceptionalism?

Ray Kurzweil

Robert Marks: Well, let’s look at what AI can do first of all. I was challenged by AI when I picked up my first calculator. That calculator could add a heck of a lot faster than I could… So all of a sudden, this electronic calculator was surpassing my capability as a human. And we know now, reading the headlines, that a computer can beat us at the world’s most difficult board game, which is go. It can beat us at chess. So yeah, there are some certain things that it can do better than us. That’s true of all technology.

But there are certain things that artificial intelligence will never achieve. These would include things like sentience, creativity, understanding — and so far, artificial intelligence doesn’t have any common sense.

Note: About creativity, no artificial intelligence has passed the Lovelace test, a measure of creativity. It is very difficult to get artificial intelligences to make common sense inferences as well. Claims that AI can “write news stories,” etc., usually depend on controlled circumstances with considerable intervention.

Gretchen Huizinga: There are people who believe that if we just have enough time, enough compute power, enough sophistication, we will get to that point. So, you’re challenging that view with this center and some of the work you’re doing?

Robert Marks: It’s very clear that in this new age, we have more data, we have greater computing power and computers are doing a lot more interesting things than they ever did before because of the availability of that data. However, there is a fundamental computer science principle called the Church–Turing thesis. The Church–Turing thesis says that the super duper computers of today… anything that they can do, you can go back to Turing’s original machine back in the 1930s, and you could do on that machine. Now, it would take you billions or trillion times as long. The thesis named after Alonzo Church and Alan Turing. Church created something called the Lambda calculus, which was a computing language. Alan Turing generated the Turing machine, the first general purpose computer. And they worked together. In fact, Turing worked under Church for a while.

Robert J Marks

And they said, “You know what? Both of these techniques do the same thing. They have the same capabilities.” And that generalized into the Church–Turing thesis.

The bottom line is that anything that the Turing machine can’t do because of fundamental physics, science, and mathematics can’t be done on today’s bigger computers.

Gretchen Huizinga: And potentially not even on the future’s bigger computers?

Robert Marks: That’s correct if they are built on the same principle that today’s computers are.

Gretchen Huizinga: I think it’s important right now to actually interject that artificial intelligence is a computer.

Robert Marks: That’s very important because yeah, anything a computer can’t do AI can’t do.

Gretchen Huizinga: No matter how many wands we wave in front of it. I want to mention that the Walter Bradley Center is part of the Discovery Institute, which has been at the forefront of scholarly research on intelligent design and challenging the hegemony of materialism and evolutionary theory that constitutes the prevailing worldview in academia. So, talk a little bit about the research that’s going on in your particular branch of this Institute and how you are tackling the fundamental difference between natural and artificial intelligence.

Robert Marks: I’ve been involved with artificial intelligence for a long time, many decades. I started to become aware of evolutionary programs purported to support Darwinian evolution. Proponents of Darwinian evolution became excited when computers were invented because they said, “We can’t reproduce this work because it would take too long to do any evolution in the lab.” Anyway, they were excited about the idea of ​​taking this algorithm, placing it on a computer and performing evolution in an accelerated sort of fashion. And they came up with a bunch of different algorithms to do it.

Now, fundamentally, Darwinian evolution is three steps: mutation of the genome, survival of the fittest, and reproduction. The idea is that, if you repeat them indefinitely, you will develop super duper things.

William Dembski

Proponents of Darwinian evolution wrote programs such as Avida. Avida is a really big one… When I came to Baylor in 2003 or so, I met design theorist William Dembski. … One of the things that we both agreed on was that evolution was unable to create information. A computer program can create no more information than an iPod can create music. It is a tool for the programmer to do something. So if these evolutionary programs were purported to generate information ex nihilo — from nothing — something was wrong.

So we developed a theory called active information, which showed that the information that was placed into these programs that predisposed them to get to the solution they wanted was infused by the programmer. And in fact, we not only argued that philosophically, we did it mathematically. We can literally measure the degree of information that is added to an evolutionary computer program that allows it to work. We published a number of papers on active information, and it has caught on. our paper has been referenced decently, picked up by some other people.

I think the big success is that we don’t see — at least in the last five, 10 years — people coming out and saying, “I have a computer program that can be creative and generate evolution.” It’s because of that work. So, I’m pretty excited and proud about that.

Next: With AI, what is hope and what is hype?

You may also wish to read about an area where AI really is making a difference — as part of much more functional prostheses for amputees. The human nervous system can work with electrical signals from complex machinery:

The Bionic Man was science fiction; the bionic hand is not. A recent internet-savvy bionic hand, developed by an American neuroscientist and computer engineer, is the most flexible yet, with sensory feedback. The trouble is, if the new bionic hands are going to help most of the world’s amputees, they can’t cost six million dollars, as in the old TV show.

Prosthetic hand controlled by thoughts alone? It’s here. Decades ago, no one could control a prosthesis only by thought. There is lots of room for the field to grow still. (2020)

New mind-controlled robot arm needs no brain implant. The thought-controlled device could help people with movement disorders control devices without the costs and risks of surgery. (2019)

High tech can help the blind see and amputees feel. It’s not a miracle; the human nervous system can work with electronic information. (2019)

Show Notes

  • 01:32 | introducing dr Robert J Marks
  • 02:38 | The Difference Between Artificial and Natural Intelligence
  • 06:31 | The Goldilock’s Position
  • 07:40 | The Challenges and Limitations to AI
  • 14:42 | The Legacy of Walter Bradley
  • 18:55 | The Difference Between Computational and Artificial Intelligence
  • 24:22 | What is Hope and What is Hype?
  • 28:44 | what keeps dr Robert J Marks Up at Night?
  • 34:26 | AI and Faith
  • 36:56 | Is Flourishing Bad and Friction Good?
  • 40:45 | The Personal Mission of Dr. Robert J Marks

Additional Resources

Podcast Transcript Download



Source link

Translate »