Artificial Intelligence - Best of Coast to Coast AM - 3/10/25

you or I or most people would define intelligence. And so within the the broad sphere of what we call AI, there are basically two different categories, and one of them is the category of of the type the type of AI that's being used for uh, digging into into complex

calculations and dealing with massive amounts of data. The traditional kinds of applications being things like you know, gigantic climate models trying to model whether it advanced or Nowadays, there's a lot of work being done in medical applications looking at X rays and other dignostic data and digging through tremendous amounts of data in ways that are difficult for humans to do, especially because these AI systems, these of this type can have the ability to find patterns pattern

matching that that aren't necessarily the best things for human beings to do, though humans still excel in lots of other areas, of course, and certain kinds of patterns humans are humans are better at. But that's the really more traditional, the good side of AI helping helping to deal with

lots of data in these kinds of ways. Now then we move over to the hype filled side of AI, and this is the side that I am very critical of, and this is mostly under the bannered generative AI, and this is where all the hype is, and this is where the big tech players, you know, Google and open

AI and the others are all hoping praying. Really, I think that this investment that they're making in it of just untold billions and billions of dollars going into building more and more data centers to crunch this data because generative AI is very consumptive of computing resources. Uh. And and in fact, they're using so much energy that this

has become a big issue now. The amount of energy they need is just extraordinary, to the point where they're trying to get permission to build their own nuclear reactors and things to power these data centers. And and this is this is the realm of the chatbots and and uh, uh you know these you know, we're all where all the hype is now, all the the the deep fakes being used to generate uh you know, fake voices and fake videos, and we're all the really controversial aspects of

AI really are now. And uh, the problem for these firms is that they've poured not only have they been pouring in all this money, but in many cases they have diverted so much of their resources, their core resources, uh they're human resources, they're engineers and and other people into AI that other aspects of their operations have have been many people would say have been neglected, and yet there is no clear sense of where the money is going to be made back from generative AI, because by

and large, the evidence so far is that while people are willing to play with it when it's free, they are much less willing to pay for it when it's not free. And so what some of the firms have been doing is just kind of bundling it in there with other stuff that people need, like Google's doing this with some of their their workplace products that are aimed at businesses, and and they just say, well, now you

get now you get the AI too. And by the way, the price of the whole packages of everything that has just gone up. And this has been very controversial because people have found in some cases their corporate information ending up in these AI systems where they didn't want it to go. And some of the problems with these systems, which are basically on what are called large language models loms, is that information that goes in can in many cases will be in a complicated manner leak out to other users.

So there's all kinds of security and privacy issues beyond the more foundational issues that so much of what these systems say tends to be incorrect, and I have a whole collection people. You know, people point out all the time these simple questions that you could ask some of these systems and they just give you completely nonsensical answers. Now, when they're nonsensical answers, you know, usually you'll be able

to recognize it. You know, if it says that three inches is equal to five inches, you know that that's wrong. And that's the kind of mistakes you actually we'll see. But the real risk of these systems is when you ask a question and the answer seems plausible, and if you're not an expert on a particular topic, you know, you know, that looks like a reasonable answer to me. But in some cases, many cases, the answer can be

completely wrong, or even more dangerous, partly wrong. And when you get an answer that's partly right and partly wrong, that's the recipe for real misinformation disinformation problems. It's sort of the heart of propaganda when you look at it from that standpoint. And it just seems like many of these problems are intrinsic to these large language models, and this is why a lot of people express concern when, for example, Google has been pushing more and more toward

using these AI overviews. There's talk now that they're going to be having versions of search. In fact, I believe they have this in the operation for some classes of users now, where the whole search experience is just talking to the talking to the AI, and you're not necessarily going to see the links the way you did, and if the answers are wrong, you're just going to have

to try to figure it out. I personally, I think it's kind of creepy when you get an answer from these things, and then in the fine print underneath it says, in essence, by the way these answers could be wrong, it's up to you to verify them. So I say, what's the point of asking a question if you're going to get an answer back? And it says it's on you to figure out if we're telling you the truth or not.

and all these kinds of things. But there doesn't seem to be any strong likelihood of those sorts of things coming into play given the way these large language models are built. I'm thinking more in terms of some completely new kind of technology that we can't really visualize now. But we couldn't have visualized a lot of technology we have now one hundred years ago, a thousand years ago, of course, So it would be foolish to say that there couldn't be some kind of intelligence technology in the

future that would be much more human like. But we're not there yet.

them for? If they're being used in a way that's you know, fraudulent, right, if you're trying, you know, if it's a deep think, if you're trying to make it sound or look like a politician is saying or doing something that they really didn't do, or anybody else for that matter. I mean, that's a real hell, that's a real problem, and governments are attempting to find ways to

deal with that through the legal system. If it's being used in ways to uh, you know, help in terms of education and positive applications, uh, that that's something else again. But there's another aspect to and this may just be my you know, my my personal preference. I generally don't like AI voices. I won't claim that I can detect them one hundred percent of the time, because they're getting pretty good, but I detect them a lot of the time.

Often there's you know, even when the voice itself is pretty good, there's off a little pacing problems or bizarre pronunciation problems, and you know, pauses and things like that, and I just I don't find it appealing at all. And actually, one of the fastest ways to get me to a boarding YouTube video is what I realize. I'm

listening to a to a to an AI voice. And Google actually is in the process of making it even easier for YouTube creators to to do this, to create AI generated videos and AI generated voices, you know, to which I kind of say to myself, you know, what to do? This is not something that I'm interested in. YouTube is a marvelous is a marvelous thing. It's a wonder of the world in many ways, but it's already saturated with a lot of often low quality AI generated

videos and more of them. Is not something that's going to be appealing to me. I'm probably at fact, I'm sure I'm not in YouTube's preferred demographic, but that's how I feel about it. Anyway.

towards streaming. And then of course that kind of opened the you know, open the floodgates, and that we have lots and lots of streaming services, only some of which will survive, some of which have already died, some of the more interesting ones there are not necessarily ones that

that most people probably even know about. But you know that that changed the whole the whole focus and and Blockbuster, you know, and I guess other if you think about you know, there were other in person video there were you know, automated systems for buying DVDs and things like that, you know, while vending machines and things.

audio over the Internet in any significant way. You know, the thought of doing video over it was was completely you know, just fantastic, But there was a lot of technology that had to be developed before it became possible. There had to be hardware, codex and a lot of advances in the hardware and the speed of these systems

to make it to make it possible. It's important to remember that when we started on the arpinet, the backbone of the internet, the high speed lines that connected fights on the arpinet where fifty six kilobits per second, that's thousand bits per second, which you know would have been a low speed modem not all that many years ago

that people would have of having their homes. So the kinds of speeds that are common now when people can get them, and a lot of people can't get them, a lot of people still can't really get decent internet at all. If that's part of our shame in this country that we can't get a good Internet to anybody, you know.

service was you could get rid of commercials. Now most of the streaming services, at least at certain tiers, have commercials and more and more of them, and of course that would drove people away from cable packages in the

first place. And now we go full circle and now it's you know, the sort of the same stuff, but now it's delivered over the internet, maybe over the same cable line that you got your your your cable TV proper and also but it's it's another one of those you know, the more things change, the more they stay the same. But you know, prices, prices will will be

going up. I really don't see. I don't think we're in a period now where we can expect to see major improvements in Internet service because there just aren't pressures

to do that. The big Internet companies are kind of sitting pretty right now, and to the extent that they don't have to do buildouts at and T has already said they're not going to do a lot of the fiber buildouts that they originally had said they wanted to do, so they have to said they're not going to bother So I think we may be entering kind of a period of stagnation in terms of that aspect of it. But the prices will stagnate, they'll they'll be going up.