I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year.
-
@cwebber Have you seen the Twilight Zone episode "The Brain Center at Whipple's"?
It's rather enlightening to answer those questions.
@socketwench I have not!
-
@socketwench I have not!
@cwebber I think it's on Tubi, although DM me if it's not.
What struck me about it is that it's from the 60s, yet the issues aren't any different compared to today.
-
(And note that *each engineer* at the org has a budget for token spend that's equivalent to what I make in a year)
@cwebber These are just bubble-maintenance costs. Way cheaper than to let the AI bubble collapse and the resulting 10-year-long recession.
-
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber
But junior engineers are made of meat, and meat might join a union, or have an independent thought, or get uppity about being owned. -
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber From management’s perspective, the problem with cheap junior engineers is that they eventually become less cheap senior engineers. Whereas the AI will be a cheap junior engineer forever
-
@cwebber From management’s perspective, the problem with cheap junior engineers is that they eventually become less cheap senior engineers. Whereas the AI will be a cheap junior engineer forever
@jalefkowit But right now the story is also "we need senior engineers to monitor the AIs"
-
@cwebber From management’s perspective, the problem with cheap junior engineers is that they eventually become less cheap senior engineers. Whereas the AI will be a cheap junior engineer forever
@jalefkowit @cwebber At least until the market consolidation sets in, at which point companies will realize that hooo boy rebuilding a talent pipeline from scratch is EXPENSIVE
-
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber Yes! this! 100% this. The junior will at least build up institutional and domain knowledge and get better every year & can use their own 🧠 to help guide choices instead of blindly doing what's asked even if it'll make things worse.
PLUS you get to feel good about giving someone insurance and a house and food.
there's NO downside.
-
@jalefkowit But right now the story is also "we need senior engineers to monitor the AIs"
@cwebber Not a problem. Senior engineers spring fully formed from the thigh of Zeus
-
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber This is one of the key thresholds I've been waiting for, when per head token spend reaches parity with engineer salary. Sounds like we're about there.
It's also been my best guess for when we might start seeing a bit more self-reflection of what we're doing. Or I'm wrong and we'll blow past it.
-
@cwebber This is one of the key thresholds I've been waiting for, when per head token spend reaches parity with engineer salary. Sounds like we're about there.
It's also been my best guess for when we might start seeing a bit more self-reflection of what we're doing. Or I'm wrong and we'll blow past it.
@dvshkn @cwebber I mean, we know the "why" here, and it's of course ghoulish.
To say what I'm sure is understood, a FTE's cost to a company is not measured by salary or benefits alone. And therein lies the value proposition for the corpos—assuming they haven't also bought into the propaganda about the models' capabilities, in which case they not only believe they are cheaper than you, but better.
-
@cwebber From management’s perspective, the problem with cheap junior engineers is that they eventually become less cheap senior engineers. Whereas the AI will be a cheap junior engineer forever
@jalefkowit @cwebber Isn't AI-as-a-service heavily subsidised at the moment though? This can't go on forever.
-
R relay@relay.publicsquare.global shared this topic
-
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber I'm not advocating replacing humans but I think the business case would be that an AI doesn't sleep, eat, take breaks, get sick, go on holiday or require a raise every year. Also an employer pays more than your salary, the line item for a human resource is salary plus payroll tax, and benefits. so it's usually your gross salary plus 30%.
I really do get where you're coming from though. AI should be a force multiplier not an excuse to layoff staff. -
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber I think you know the answer here, because it's not that they're paying for code to be produced; the CTO is paying for bragging rights so he can proudly trumpet their adoption numbers to anyone within earshot at all the cool CTO parties he goes to
-
I have a friend who has a budget where she spends as much individually on AI-as-a-service tokens as I make in a year. And it's acknowledged that the system misbehaves, needs to be monitored closely like a junior engineer, etc.
So why not hire some junior engineers if you're an org that has that equivalent cash to spend? Companies that are in such a position: you've never had a better market chance to get a sweet deal on young talent
@cwebber because that would mean spending money on wages and companies are forbidden by their religion to do so
-
@jalefkowit @cwebber Isn't AI-as-a-service heavily subsidised at the moment though? This can't go on forever.
@dukeboitans @jalefkowit @cwebber no you need to understand, the genius is that taxpayer money the AI companies get is partially spent to line the pockets of the people deciding on subsidies. they can keep that up for as long as it takes
-
@cwebber because that would mean spending money on wages and companies are forbidden by their religion to do so
-
(And note that *each engineer* at the org has a budget for token spend that's equivalent to what I make in a year)
@cwebber i bet that's completely sustainable and will last forever
-
@cwebber I'm not advocating replacing humans but I think the business case would be that an AI doesn't sleep, eat, take breaks, get sick, go on holiday or require a raise every year. Also an employer pays more than your salary, the line item for a human resource is salary plus payroll tax, and benefits. so it's usually your gross salary plus 30%.
I really do get where you're coming from though. AI should be a force multiplier not an excuse to layoff staff.@mike Is it true that AI won't require a raise every year? I wouldn't be surprised that executives would think that, but it seems like token price inflation would be a lot harder for companies to mitigate than it has been for them to suppress wages. AI has the whole multi-billion dollar company apparatus behind it, most employees are lucky if they can afford an employment lawyer to read their contracts.
-
@mike Is it true that AI won't require a raise every year? I wouldn't be surprised that executives would think that, but it seems like token price inflation would be a lot harder for companies to mitigate than it has been for them to suppress wages. AI has the whole multi-billion dollar company apparatus behind it, most employees are lucky if they can afford an employment lawyer to read their contracts.
@cargot_robbie @cwebber Fair point
