Right, but in the same spirit, we’re not just talking about LLMs. If we’re being accurate, the common interface we’re used to with ChatGPT or Gemini, they are a system of different models including LLMs and other models for images, or sound.
If we’re talking AI in movies and music, besides LLMs for writing, we’re mainly concerned with diffusion models
If we’re talking AI for wearables… That’s usually more on the sensor/classification side of ML, so it’s not even generative.
“Big Data”, “Neural Net”, etc etc, they keep changing the buzzword for this crap for the past decade.
AI just resonates with the public because everyone has seen Terminator movies. Your dipshit CEO of Buttplugs Inc has no clue what an LLM, or neural net is. But he knows what AI is and he delusionally believes it’s going make Buttplug inc. more profitable in the future.
They call it a machine learning algorithm i call it lossy text compression.
When you give it a prompt it uncompresses a small portion of the petabytes of zipped information based on the relevance of the keywords you choose. The “generation” comes from approximations of keywords and the underlying prompt. It isn’t “generating” information as much as big tech wants you to believe, its taking information from multiple sources and combining it into one contiguously flowing statement. For people who didn’t understand what I said, generative AI operates similarly by taking 2 paint colors and making a new kind of color from it. It isn’t critical thinking, it’s more of experimenting with no basis, reason, or hypothesis.
To conclude, the underlying mechanism is a lossy compression algorithm: there is no way to reduce the size from petabytes of information to a sheer 40gb downloadable size besides through lossy compression. The “generative AI” component is analogisly equivalent to making a new color of paint by combining two or more other colored paints. It’s entire existence is through experimentation with no preamtive basis that adding two or more colors would give you a new color. It will never figure that it even after having done the experiment.
Yeah, for all the undeserved hype about AI that we and I’m sure plenty of C-suites know about, there is still the issue of fiduciary duty to their bosses (shareholders).
If you’re a tech CEO that knows it’s all crap, but the market conditions are such that going all-in on AI is going to triple your share price for no good reason, you’ll probably be in a “get that money or we’ll find somebody who will” situation.
But that’s also why executives get so much of their compensation in the form of stock. There’s no need for shareholders to get off their asses to demand short term gains they can cash in on when the decision makers are one of them and playing the game game.
if hover boards were legit real you’d have so many injuries the company would be sued into insolvency.
already starting to see this with ebikes… massive increases in head trauma numbers since they became popular. mostly from old people buying a $5000 e bike and then riding it w/o a helmet at 15-20mph.
I’m going to be that guy and point out that LLM’s are not really “AI”, that’s just the corporate buzzword but “AI” is a loosely defined thing.
I think we should all get better at calling them LLM’s publicly to take some of the magic woo-woo away.
The term “AI” today is almost like the term “computer” decades ago.
As in, “this new vehicle is more efficient than ever thanks to a new aerodynamic shape created not on the drawing board, but on the computer.”
Ha! Yes so true.
Right, but in the same spirit, we’re not just talking about LLMs. If we’re being accurate, the common interface we’re used to with ChatGPT or Gemini, they are a system of different models including LLMs and other models for images, or sound.
If we’re talking AI in movies and music, besides LLMs for writing, we’re mainly concerned with diffusion models
If we’re talking AI for wearables… That’s usually more on the sensor/classification side of ML, so it’s not even generative.
I think we should call it with the more appropriate acronym: Systematic Approaches to Learning Algorithms and Machine Inferences (SALAMI)
Then they can fight over who has the biggest SALAMI
I agreed years ago, but it’s too late now.
“Big Data”, “Neural Net”, etc etc, they keep changing the buzzword for this crap for the past decade.
AI just resonates with the public because everyone has seen Terminator movies. Your dipshit CEO of Buttplugs Inc has no clue what an LLM, or neural net is. But he knows what AI is and he delusionally believes it’s going make Buttplug inc. more profitable in the future.
It’s effective marketing.
They call it a machine learning algorithm i call it lossy text compression.
When you give it a prompt it uncompresses a small portion of the petabytes of zipped information based on the relevance of the keywords you choose. The “generation” comes from approximations of keywords and the underlying prompt. It isn’t “generating” information as much as big tech wants you to believe, its taking information from multiple sources and combining it into one contiguously flowing statement. For people who didn’t understand what I said, generative AI operates similarly by taking 2 paint colors and making a new kind of color from it. It isn’t critical thinking, it’s more of experimenting with no basis, reason, or hypothesis.
To conclude, the underlying mechanism is a lossy compression algorithm: there is no way to reduce the size from petabytes of information to a sheer 40gb downloadable size besides through lossy compression. The “generative AI” component is analogisly equivalent to making a new color of paint by combining two or more other colored paints. It’s entire existence is through experimentation with no preamtive basis that adding two or more colors would give you a new color. It will never figure that it even after having done the experiment.
Yeah, for all the undeserved hype about AI that we and I’m sure plenty of C-suites know about, there is still the issue of fiduciary duty to their bosses (shareholders).
If you’re a tech CEO that knows it’s all crap, but the market conditions are such that going all-in on AI is going to triple your share price for no good reason, you’ll probably be in a “get that money or we’ll find somebody who will” situation.
But that’s also why executives get so much of their compensation in the form of stock. There’s no need for shareholders to get off their asses to demand short term gains they can cash in on when the decision makers are one of them and playing the game game.
i saw my first hover board in back to the future, the pink one. now it’s that overpriced exploding segway.
at least the name segway was almost creative
Remember when the segway was a mysterious to-be-revealed “it” that we were going to design our cities around in the future?
I 'member.
if hover boards were legit real you’d have so many injuries the company would be sued into insolvency.
already starting to see this with ebikes… massive increases in head trauma numbers since they became popular. mostly from old people buying a $5000 e bike and then riding it w/o a helmet at 15-20mph.
I’m so sad we’re not seeing this with AI. The injuries we get, the insolvency not so much. Fuck.
that didn’t stop RAZOR scooters
yes, I’ll call them AI when they are sentient D:
I was spoiled by Iain Banks’s Culture <3
Iain M. Banks, don’t ya know!?!?
I’m still saving a culture book or two to enjoy, but the entire setting has been amazing.
Eh, afaik he only added the letter to differentiate his scifi books from his non-scifi books :P
Yes, exactly!
Yeah, it’s M for the Culture stuff