Well, they’re trying to solve work scarcity. I’d argue reading that as “wages” is an inherently capitalist take.
Mind you, they are not succeeding at fixing work scarcity, so the point is kinda moot. “AI will take your job” is the magic centre of the Venn diagram where AI shills and AI haters overlap.
Hah. Hey, I’m not even saying the tech is useless, but best case scenario that’s our PhD student friend using ML to process data faster, or in ways that weren’t feasible before, not being replaced by an AI PhD student.
20 years ago, we had 9 people behind the camera running a live local newscast (Floor Director, Cam Operator, Teleprompter, Chyron, Graphics, Video Playback, Live/Commercial Cut-in, Audio, and Director). Now, in a market three times the size, the same job is done with 3 people and a metric ton of automation. What used to feel like a bridge crew piloting a ship now feels like conducting corpo bots within time-frames that prevent giving any of them real attention. I do believe most AI systems will continue to need people in the loop. It’ll just be fewer people in less fulfilling positions.
OK, not disputing that, but that process has eff all to do with AI. Gen AI gave people a recognizable target, but automation was done using good old dumb algorithms for a long time before we taught computers to babble like a toddler. I was in the room for a ton of “can we automate all this QA” when machine learning was failing to tell a cat apart from a bycicle.
Also, for your specific case I think Youtube and social media had a TON to do with the shifting standards of running a skeleton crew TV studio. Ditto for the press in general. Remember when copy editors were a thing?
I don’t even know what it can do that’s useful. The prompter maybe, and captioning if you’re feeling frisky and don’t mind airing something insane by accident.
But what, you’re going to let an AI handle chyrons and cut-ins? I did briefly work at a TV station and back then we had two separate continuity guys and three redundant automated sources for all canned content just to make sure you never got a black frame. I once saw a guy get fired for three seconds of dead air in a commercial break because at least back then absolutely any mistake around commercials was a huge, automatic money loss.
I absolutely believe you when you say it’s degraded, because… well, again, Youtube and Netflix, but at most you can… you know, cut one of the two guys so you can still fire the other when the AI plugs in something random instead of an ad.
Alright, let me rephrase that. You can definitely cut more than that, but you’re probably going to have to un-cut that pretty fast when some AI claims that someone is an international art thief in a chyron or something.
Somehow, all AI manages to do is strip the innovation and creativity out of the most exciting career fields.
The rote physical labor of polishing the end product, marketing it to the masses, and distributing it via service sector retail facilities seems to stubbornly persist.
Well, I don’t know about that. I mean, I haven’t integrated any AI in my personal workflow at all beyond… I don’t know, maybe not remembering something and finding that faster than a classic search engine just to remember the name.
But in the places around me where I do hear people picking bits of it up I see it used for what? Proofreading and rote, repetitive tasks? I don’t know that it’s productive at all for even that, beyond expensive, custom-trained ML processes that have little to do with commercial generative AI.
not remembering something and finding that faster than a classic search engine
That’s more a consequence of Google Search capitulating to the ad sales side of the business at the expense of search efficency. Same thing happened to Yahoo and Lexus Nexus.
where I do hear people picking bits of it up I see it used for what? Proofreading and rote, repetitive tasks? I don’t know that it’s productive at all for even that, beyond expensive, custom-trained ML processes
Amazon has heavily invested in generative AI for its screenwriting and book sales business. Consequently, their original programming has suffered and their book marketplace flooded with crap.
No, I don’t think that’s the case. For one, I don’t use google for search, I’m not an animal.
But for another, I don’t use AI search to replace classic search, I only use it when a) I already know the answer but I can’t remember it, and b) the query is so fuzzy it’d take too long to refine on classic search. Think of “hey, what was the name of that movie where the Home Alone kid was with Frodo Baggins and one of them was nuts?”
Incidentally, I just tried asking that to ChatGPT and it got it right.
As for the other thing, I don’t know if that’s accurate, but if it were, it’d be exactly what I’m talking about. Not saying people won’t try, but if and when they do, they’ll learn pretty quickly that it’s a bad idea.
I don’t use AI search to replace classic search, I only use it when a) I already know the answer but I can’t remember it, and b) the query is so fuzzy it’d take too long to refine on classic search.
Google used to bill its search software as high quality artificial intelligence capable of returning useful answers to fuzzy questions and reliable responses to repeated inquiries. Only recently has the search engine prioritized “new” information over reliable sources and begun aggressively injecting ads into every search.
Modern AI is nice because its not overflowing with Ads and it does appear to weight the results by usefulness rather than newness. But how long do we expect that to last in a market where consistency and clarity are at odds with revenue generation?
Well, it could bill whatever it wanted, and it was pretty good at parsing queries, but it was all smart programming over dumb code breaking down whatever you wrote. It certainly couldn’t handle natural language and fuzzy requests particularly well.
BUT the flipside of that is that, ads or no ads, you can’t trust gen AI results at all. Which means you should never, EVER ask gen AI any question you don’t already know the answer to or aren’t willing to verify.
And if you’re going to verify it (and potentially learn it’s wrong and research it all over again the classic way) you are now taking longer to get the same answer with AI.
It’s getting worse the more it relies on being a parser for classic search, too. Anything that isn’t page 1 results on Google or Bing it just won’t acknowledge, so the worse classic search gets, the worse newer AI search gets, too.
I genuinely thought that would be a good application when they first came up with AI chatbots, but… yeah, no, I was wrong. At least outside the specific use case I outlined above.
Well, they’re trying to solve work scarcity. I’d argue reading that as “wages” is an inherently capitalist take.
Mind you, they are not succeeding at fixing work scarcity, so the point is kinda moot. “AI will take your job” is the magic centre of the Venn diagram where AI shills and AI haters overlap.
Somewhere, a PhD student 2 years into research on a single protein structure raises an eyebrow.
Hah. Hey, I’m not even saying the tech is useless, but best case scenario that’s our PhD student friend using ML to process data faster, or in ways that weren’t feasible before, not being replaced by an AI PhD student.
20 years ago, we had 9 people behind the camera running a live local newscast (Floor Director, Cam Operator, Teleprompter, Chyron, Graphics, Video Playback, Live/Commercial Cut-in, Audio, and Director). Now, in a market three times the size, the same job is done with 3 people and a metric ton of automation. What used to feel like a bridge crew piloting a ship now feels like conducting corpo bots within time-frames that prevent giving any of them real attention. I do believe most AI systems will continue to need people in the loop. It’ll just be fewer people in less fulfilling positions.
Citing the same time period, it used to be each local station had a Master Control Operator.
Now an MCO is expected to run 10 stations all at once from a remote location. No change in pay. Just more responsibility.
OK, not disputing that, but that process has eff all to do with AI. Gen AI gave people a recognizable target, but automation was done using good old dumb algorithms for a long time before we taught computers to babble like a toddler. I was in the room for a ton of “can we automate all this QA” when machine learning was failing to tell a cat apart from a bycicle.
Also, for your specific case I think Youtube and social media had a TON to do with the shifting standards of running a skeleton crew TV studio. Ditto for the press in general. Remember when copy editors were a thing?
YouTube and Social Media were part of the '05 (algorithmic) AI wave, yeah.
I don’t even know what it can do that’s useful. The prompter maybe, and captioning if you’re feeling frisky and don’t mind airing something insane by accident.
But what, you’re going to let an AI handle chyrons and cut-ins? I did briefly work at a TV station and back then we had two separate continuity guys and three redundant automated sources for all canned content just to make sure you never got a black frame. I once saw a guy get fired for three seconds of dead air in a commercial break because at least back then absolutely any mistake around commercials was a huge, automatic money loss.
I absolutely believe you when you say it’s degraded, because… well, again, Youtube and Netflix, but at most you can… you know, cut one of the two guys so you can still fire the other when the AI plugs in something random instead of an ad.
Alright, let me rephrase that. You can definitely cut more than that, but you’re probably going to have to un-cut that pretty fast when some AI claims that someone is an international art thief in a chyron or something.
Somehow, all AI manages to do is strip the innovation and creativity out of the most exciting career fields.
The rote physical labor of polishing the end product, marketing it to the masses, and distributing it via service sector retail facilities seems to stubbornly persist.
Well, I don’t know about that. I mean, I haven’t integrated any AI in my personal workflow at all beyond… I don’t know, maybe not remembering something and finding that faster than a classic search engine just to remember the name.
But in the places around me where I do hear people picking bits of it up I see it used for what? Proofreading and rote, repetitive tasks? I don’t know that it’s productive at all for even that, beyond expensive, custom-trained ML processes that have little to do with commercial generative AI.
That’s more a consequence of Google Search capitulating to the ad sales side of the business at the expense of search efficency. Same thing happened to Yahoo and Lexus Nexus.
Amazon has heavily invested in generative AI for its screenwriting and book sales business. Consequently, their original programming has suffered and their book marketplace flooded with crap.
No, I don’t think that’s the case. For one, I don’t use google for search, I’m not an animal.
But for another, I don’t use AI search to replace classic search, I only use it when a) I already know the answer but I can’t remember it, and b) the query is so fuzzy it’d take too long to refine on classic search. Think of “hey, what was the name of that movie where the Home Alone kid was with Frodo Baggins and one of them was nuts?”
Incidentally, I just tried asking that to ChatGPT and it got it right.
As for the other thing, I don’t know if that’s accurate, but if it were, it’d be exactly what I’m talking about. Not saying people won’t try, but if and when they do, they’ll learn pretty quickly that it’s a bad idea.
Google used to bill its search software as high quality artificial intelligence capable of returning useful answers to fuzzy questions and reliable responses to repeated inquiries. Only recently has the search engine prioritized “new” information over reliable sources and begun aggressively injecting ads into every search.
Modern AI is nice because its not overflowing with Ads and it does appear to weight the results by usefulness rather than newness. But how long do we expect that to last in a market where consistency and clarity are at odds with revenue generation?
Well, it could bill whatever it wanted, and it was pretty good at parsing queries, but it was all smart programming over dumb code breaking down whatever you wrote. It certainly couldn’t handle natural language and fuzzy requests particularly well.
BUT the flipside of that is that, ads or no ads, you can’t trust gen AI results at all. Which means you should never, EVER ask gen AI any question you don’t already know the answer to or aren’t willing to verify.
And if you’re going to verify it (and potentially learn it’s wrong and research it all over again the classic way) you are now taking longer to get the same answer with AI.
It’s getting worse the more it relies on being a parser for classic search, too. Anything that isn’t page 1 results on Google or Bing it just won’t acknowledge, so the worse classic search gets, the worse newer AI search gets, too.
I genuinely thought that would be a good application when they first came up with AI chatbots, but… yeah, no, I was wrong. At least outside the specific use case I outlined above.