Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
AI and the Flimflam Man
Courtesy of NightCafe. An AI-drawn mentalist, though more fantastical than I had in mind.
To the amazement of the crowd nearly one third of them stand up. “I knew it!” shouts the mentalist, “but I almost didn’t pick 37; I thought of 39 at first. How many of you picked 39?” Over half the audience is now standing, to wild applause.
I think I pulled that example from Michael Crichton’s State of Fear, but it’s been so long since I’ve read the book that I’m no longer sure. Let’s figure out the trick. If I asked you to guess a number between one and 50, fair odds say you’d get it right 2% of the time. The mentalist does a great deal better than that. How? He isn’t asking for a number between one and fifty.
Since the number has to be two digits we lose [1,2,3,4,5,6,7,8,9] right away. Since both of the digits must be odd we lose everything in the 20s and 40s too. Not quite; there’s going to be some guy in the audience fervently thinking “29” who’s forgotten that requirement, but leave him aside. Since both digits are odd you only get half of the decades that are left. Here’s all remaining possibilities:
[11, 13, 15, 17, 19, 31, 33, 35, 37, 39]
That’s only ten numbers; the odds are way better already. From there you have to go into psychology. People are less likely to choose a number with a repeated digit [11, 33], and perhaps the kind of folks who believe already will shy away from an unlucky number [13]. After that (and I’m afraid I’ve forgotten the specifics) people pick some numbers preferentially. Survey enough people and you get the idea that 37 stands out. Between that and how the mentalist took credit for two numbers you can see how he’d get half the room.
The same picture as the first, only a craggy mountain valley.
What does that all have to do with AI? Whether it’s a stage mentalist or a phishing email or a pretty lady in a bar who just happens to come from your home town, the way to suss out the scheme is to watch inputs and outputs. You’ll hear a lot of talk about the middle — how the machine learning algorithms rely on matrices full of numbers that not even their programmers can decode, or the mystical wonder of telepathy, or how the quants in the basement have the stock market beat — but keep a very sharp look at those inputs. A flimflam man will pick up on a signal you didn’t realize you sent and amplify it right back to you as a successful prediction.
That’s why I’m not worried about AI. It isn’t a scam, but think about it in that model for a minute. When I went to make that picture up top, I typed in a description of what I wanted to see; an excited mentalist on stage. That’s the input. The AI didn’t have the genius to come up with the idea itself; it didn’t pluck a particle of inspiration out of the ether. Want proof? Take a look at the second drawing. It’s the same one as the first, only I changed the prompt to “Show me something you want to draw.” That’s nonsense to an AI. It can’t understand what I asked, so it gives back essentially a nonsense answer. It drew the same scene, only as a mountain valley.
Now outputs. The input is you tell it what picture to draw, the output is you tell it what picture you’ll accept. The AI took my first prompt and decided that I wanted a fantastical scene with actual magic going on. The first picture was so interested in the crowd that it crowded out the mentalist. I amended the prompt to “Excited mentalist on stage, with focus on the mentalist.” I was hoping for more of a stage magician approach, but the picture is cool enough to share.
AI isn’t a scam. There’s magic going on inside that black box even if you have to provide some yourself. Whether or not those pictures are what exactly I had in mind, they’re better than I could create myself, at least without intensive training. To the extent that this post could be written by ChatGPT it’s because the machine has successfully duplicated all the rules of composition and grammar that I only imperfectly understand. Machine Learning is a powerful tool, and I look forward with hope for it to automate away the drudgery in a great many jobs.
But remember those inputs and outputs. By prompting the AI with your idea, and by selecting an acceptable execution of that idea, you’re providing the creativity that the machine can’t automate. The machines will make it easier to replace art. They will not replace art as a profession. The robots can now churn out pulp novels. They won’t make the next great work of fiction. It might be that your secure job is now threatened. Your job was never secure from the future. It’s only con men who told you otherwise.
Published in Science & Technology
What’s going to happen when the samples that AI train on contain a significant fraction of AI generated art?
I’m going to assume that it ends up as the clone of a clone problem until proven otherwise.
Samuel Johnson said of a dog standing on it’s hind legs that the wonder of it isn’t that he does it well, it’s that he does it at all. Can you imagine how astounded men were by Edison’s phonograph, and can you imagine how crackly and staticky the replayed music sounds to modern ears? Flash forward a hundred years, the reproduction is prefect, but we still go out to live music. Why? There’s something in live music the mechanical reproduction doesn’t convey.
What happens when the same is true for the artists?
Yeah, there are girls dancing there.
“The machines will make it easier to replace art. They will not replace art as a profession. The robots can now churn out pulp novels. They won’t make the next great work of fiction.”
Many of those “great works of fiction” were just the popular or “pulp” fiction of the day. I truly enjoy Dickens and so did the masses in the mid 19th century. I agree that AI might not replicate what the critics consider great fiction today…because it’s dreck and I expect the AI will likely do better and supply an actual story.
A truly great post, from Hank Rhody, the one-man Great Post Factory.
Plus, continuing the theme: no kidding, the moment you referred to the too-friendly girl at the bar, I just knew what clip would come up, and it did! What are the chances that Hank and I would both be thinking of the same moment in the same TV show that aired when I was four, and Hank wouldn’t be born for another 30 years?
So let’s apply some of the deductive power of the OP. It’s probably not totally at random, since somewhere in the back of Hank’s rather well-stocked mind may be occasional mentions of Sgt. Bilko on R> threads.
I have many times been disappointed by the live music I experienced in comparison to the production experienced on vinyl, 8-Track, cassette, cd, or the various different file types I utilized on computers. It is sublime when the live experience surpasses them but it is distinctly in the minority.
That’s true insofar as they’re written not so much because the authors thought they’d be adding to the tradition of Shakespeare and Milton, but there’s a qualitative difference between what they put out and what the other pulp writers did. Robert Heinlein was writing explicitly to make a buck, but even reading him in his historical context, next to other writers of the same era and genre, you can tell there’s something in his stuff that wasn’t in a Malcom Jameson or a Jake Hawkins story.
The good stuff survives longer because it’s better.
That episode really is impressive for a number of reasons. One of them because it so clearly demonstrates the nature of a con.
Another part of it is that what people hear on their phones etc, is far from prefect. Not to mention far from perfect. You can get the full experience at a concert hall etc, getting it at home is more of a problem than most people want to deal with.
A lot of music is the bass, and no matter how much you adjust your EQ settings, you simply cannot get proper bass from a phone, or ear buds… or even high-end headphones. Because bass is for your whole body, not just your ears. That’s actually true of music in general, but especially bass.
When I did car stereo installations after High School at a high-end audio shop in Oregon, I got to hang around some of the really good stuff. Yes, it was 40 years ago, but no matter what you think, your iPod DOES NOT sound better! Even if you have “Beats By Dr Dre” headphones or something. Maybe you think it sounds better than what you DID have before, but it’s not better than what you COULD have, if you really meant it.
In particular, there was a Harman-Kardon rackmount setup with GLi speakers that were about chest-high. When playing Steve Miller Band’s “Swing Town,” you could feel your shirt start to ripple when the TRUMPETS came in!
Also check out my comment at:
https://ricochet.com/948695/jim-steinman-rip/#comment-5430641
My current living-room setup, the upper speakers on the wall-mounts are Polk RM-3000 series with a passive subwoofer behind the cabinet. Below those are Polk RM-8 front surround speakers that also go with a front center speaker plus RM-7 speakers for rear surround, and a PSW-10 powered subwoofer. (I haven’t set up my fully-LOUD system yet, which is Mitsubishi dual-mono components and Polk “bookshelf” speakers on stands plus another powered subwoofer. I also have a set of Laboratory Reference Series – LRS – components from Luxman, vintage 1978, but that’s a more “elegant” setup.)
(This photo is from last year, things have cleaned up some since then.)
Confessions of an audiophile aside, I maintain even with a high end home setup there’s something different about live music than canned.
It’s still not perfect – or even prefect – but you can get a lot closer than people can with an iPod etc. At least in audio quality. You can’t – at least if you’re lucky – quite duplicate the smells, etc.
And it’s also true that people don’t necessarily go to a live show particularly for the music anyway. Certainly not if they’re trying to understand the lyrics, etc.
I’m more and more convinced that machines are “composing” what passes for music anymore.
I’ve elsewhere opined that taking bits of writing from everywhere and pasting them into a new version which doesn’t really add anything new is what hack writers do. Now I’m thinking about music creation . Take the “AI” black box away and plug in a trained musician. Now add the same inputs and outputs. “This is what I want.” “No, not like that; make it jazzier.”
How would you distinguish AI generated songs from executive driven music that’s been focus grouped into oblivion?
In some ways, what AI seems to do – at least so far – may be the ultimate focus group.
It’s my wife’s favorite episode, because it shows Bilko using his superpowers (lying and crooked gambling) in the noble service of good, fleecing a couple of less likeable con men. Also, since we all know Ernie going in, there’s built-in laughs in his innocent routine.
Still, the Bilko reference seems like a remarkable coincidence, one of these what-are-the-odds deals. I am thinking of a scene with the original cast of Mission: Impossible. What do I see?
“Cinnamon Carter as the fake daughter of a Nazi big shot, holding a copy of Mein Kampf, saying “Zere iss alvays zomething elss to learn”.
That is correct.
It wouldn’t be tough to write a program to synthetically ape Lukasz Gottwald and Max Martin.
An excellent post with a great conclusion.
I believe you are correct, that the creative spark, the flash of insight, the amazing melody, and whatever else is something only a human can create. Or at least a non Turing machine entity.
I agree with Roger Penrose that consciousness is not computable. And our brains, our minds, do things that are not computable.
Not much. Computers can process things in a nanosecond and all the tricks they deploy post in studio they now execute live at venues. A sports audio guy I know told the story of seeing a very famous group on tour and looked at the mixer in the middle of the arena and realized he knew the guy. He went over just to say hello and his friend invited him on to the platform. There he handed him a set a headphones and said, “Here, you want to know what these guys really sound like?”
It was the awful, most out-of-key singing and playing he had ever heard. But a bevy of software put everything back in pitch.
I’ve heard of auto-tune for singers, of course, but how could an entire band be processed that way? Do they somehow use a previous recording running at the same time?
Sure it could. Each instrument and each singer is coming in on a separate input, and each is run through pitch-bending. (You don’t have to bother with the drums, which may be anywhere from three to seven microphones all by themselves. )
Back in my day as a recording engineer, devices like mutrons and dynamic range compression were expensive enough that there were only a few available at the mixing board, not one on every channel of input. You had to choose what got the effect. Now, the sky’s the limit.
That looks like a pretty righteous stereo rig, K. Back in the day, the speakers were JBL, Rectilinear, Altec or Klipschorn.
A couple of my friends were hi-fi snobs with amps by Crown and McIntosh Labs. They referred to Harman-Kardon as “Hardon-Cardboard”.
A lot of HK equipment I didn’t care much for, especially when they got into plastic-front receivers etc. They seemed to somehow break down at times even when turned off. But the Citation 16 power amp and matching preamp – Citation 11, I think – were pretty good. And that’s what was connected to the GLi speakers.
I was never very impressed with Crown or McIntosh especially considering how much they cost. Klipsch had the same problem, in addition to not being very good at lower volume levels. Other than possibly the huge Corner Horn models – which I have heard – no Klipsch model no matter how expensive can have the bass response you’ll get from a good subwoofer.
I’ve had people listen to my setups who wonder “where’s the bass?” because it’s not going BOOM BOOM BOOM! like their car stereo or their portable boom-box or whatever. But then I take them outside and they notice the windows and walls vibrating.
Another sad fact is that many people don’t have such great hearing to start with, especially if they’ve spent much time in very loud clubs or cars, etc. Along with people who just didn’t seem to have very good hearing from birth. My first college roommate couldn’t hear the “whicka-whicka” sounds in Donna Summers’ “I Feel Love” so he didn’t realize that his DFS speakers didn’t even reproduce them, the way my Evolution speakers (same foundation, but with a better tweeter and different crossover) did.
And I’ve done side-by-side comparisons with many different speaker brands and models. It seems to be unique to the RM-3000 style somehow, because even other Polk RM types don’t match them. When I’ve had other high-end speakers – including other Polk models – and the RMs compared side-by-side, the other speakers sound fine but I can still tell that speakers are present. When I switch to the RMs, the physical speakers seem to disappear and there’s just music.
Hank’s post is s reminder that many developments are correctly anticipated but seem to take forever to happen. AI was expected to be conscious by 2001. Then for decades, that was a sad joke; it wasn’t going anywhere. The leading self-driving robot of the Seventies, “Shakey” (sp), took two hours to cross a room. Nuclear fusion was thought to be easy…easy…aw, forget it, it’s almost impossible.
And now, here we are, once again on the brink of huge changes. It becomes tough to define what I don’t think we’ll see in the foreseeable future. My own list is, time travel; radio transmission of matter; faster than light travel.
That depends on how far you think is “foreseeable” of course. I suspect that time travel may be ultimately impossible, at best we might be able to “view” the past but not actually go there since that might allow for changing what has already happened. “Viewing” the future might be either impossible or pointless.
FTL travel may be possible but only outside of a “gravity well” such as a star or even a planet, which means it would be impossible to discover or even really experiment until we can get beyond or at least much closer to the edge of the solar system.
“Radio transmission of matter” seems pretty much an oxymoron, and even if some mechanism were developed to scan something at one end and basically 3d-print a duplicate at the other end, it wouldn’t be the same item.
Seven? Seven? Make ’em set up in the hallway outside the studio. They’re too loud as it is.
A rigorous, mathematical basis for psychology. Was just reading an old science fiction story about that. Or rather, one that assumed such a thing would have been worked out by now.
Yes, fluctuating hormones momentarily coursing in our blood streams due to emotional responses change our thinking processes.
Heinlein wrote about controlling the masses through carefully calculated messages (The Roads Must Roll, I think).
I think that Google, Facebook, et al are getting there.