Post

The kind of AI that looks at X-rays and tells you if you’re about to suffer from lung cancer is good. The kind that hallucinates and encourages bad habits because it’s been told to is bad. Unfortunately C-Suite marketing execs don’t understand the difference.

Maybe you could revisit “The Moon Is A Harsh Mistress” as you did with “Little Fuzzy”?

It's pretty unregulated here and they are popping up everywhere. Unregulated. And chasing super intelligence because they think other countries are, is just asinine. They are basically building a giant machine coffin at this point.

Call me old fashioned, but my #1 concern about “AI” is that “SkyNet” will become a reality. That the geniuses currently in charge at the top will conclude that the best way to handle a swarm of 21 century threats will be to let AI handle bc, you know, it makes decisions much faster than us humans.

for me, tech gradually started switching from releasing dopamine to spiking cortisol around the time FAANG blew up. i went from an enthusiastic early adopter to feeling a sense of existential dread every time a techbro opens his word hole.

If we don't exterminate ourselves first, eventually regulation will coalesce, appropriate use patterns will emerge, and it will become just another tool in the belt of the human race.

Agreed. It’s like no one can remember GIGO still applies anymore. Except the musky types like it that way…

And that was assuming no interrupts or crashes. Division head backed off, realizing there was no opportunity to cut headcount, meaning the engineer employed to calculate weights. This was for aerospace usage, where weights are critical to flight/lift.

100% agree - worst trouble we had in engineering was salesman promised the impossible to our customers, THEN asked if we could do it. Salesman told division head software could "calculate weights at the push of a button." We explained 3D model & surfacing every nut/bolt & system calc slow for 3 wks

But thinking is HARD! 😭

Remember when Apple Maps first came out and got dragged because it sometimes gave a bad route or didn’t know where something was? We are now supposed to base our entire economy on an even more unreliable technology

I'd rather they just used AI to scan my test results/screening images and leave the creativity angle to folks who understand the number of fingers on a hand and, well, physics on a basic level.

or on the upside, let's hype it up, so when 'the whole economy's going to crash' we have something, the "uncontrollable AI", to blame, conveniently no further questions will be asked ... ;-)

if math were decided by committee …

Recently, I was shopping for a new modem, and I submitted my requirements to several different AIs, asking for the results in a spreadsheet. All of them returned something, but… - widely varying numbers from 8 to 29. - Some generated a spreadsheet, others a csv, and others just raw data. -

1. Opposite of tech nerd = me: a technological philistine; not a choice but inherent aversion & seemingly irrational inability to understand basic type of information that millions of people not more intelligent or educated (“less intelligent” sounds so OBNOXIOUS!) have NO PROBLEM comprehending!

Yep. We've got leaders desperately trying to preserve a bubble.

and it’s soooo inefficient!

"Follow the money." It's always about rich people making themselves richer.

I can’t find the skeet anymore, but someone posted recently: Technology used to be tool I could use to solve problems and my life better. Now it’s something being forced on me against my will.

As a software developer, I’ve found that a good use case is building test data, especially for volume testing. It’s a simple but boring task, and AI can generate a lot of test data in a short time. When the tasks get more complex, you start to get more hallucinations…

And you absolutely have a program only trained on one type of data and have humans very thoughtful about the training data to avoid biasing it because the 'abnormal' data set has other differences from the 'normal' data set.

I've read sci-fi, we have to go thru the AI turns evil and all humans must die phase before we get to the friendly helpful AI phase. But does it have to happen in my lifetime?

And those dot-com startups were infinitely realer than genAI ...

Somebody slapped a public chatbot interface on a promising tech that was still largely in its infancy for practical/enterprise purposes, at a moment juuuust close enough to crypto crash, when techbros and grifters NEEDED a new buzzword to drain Dumb Money... and now here we are.

They need another thing to make number go up. There are no new real innovations in tech they can tout, so they need to make this one work, and now that the sunk cost fallacy has wormed its way into their Business Idiot brains, we're stuck until the money implodes.

Even more than the cryptocurrencies that preceded it, AI is an excuse to build hyperscaled data centres that demand energy, preferably of the fossil variety.

It doesn't scale. They haven't figured out how to make it pay. They are just delaying the inevitable.

I'd always thought that science and technology were about providing new ways to view the world, and do things -- not about settling on one (often sub-par) strategy and doubling/tripling down on it, while ignoring all criticism and other suggestions for fixing problems.

I has to use Chat at a job to make furniture fabric all the same color for a website. It always took liberties in changing the design unprompted and the rooms it did were always off. I wasn’t a fan.

I've heard it compared to how radium was in everything, for a brief while.

I think that's two different things. The companies have charged forward, huffing their own farts like it's nitrous oxide, assuming the entire world would be as enthusiastic as they are. The money will come when people catch up to them - and they all want to be the only option when that happens.

But I am just a bitter old software engineer who talked his mouth off warning about the obvious consequences of this inanity. Call me Cassandra :(

“You drink it”

Just have to remember that while being dangerously bad at most things, it is _also_ dangerously good at some things, and that blade is very jagged

When people used the internet - they couldn't stop. When they got the iPhone - they couldn't stop. Even when they got their first computer - they gravitated toward it. It feels like the people trying to sell it to us have to find ways to insist we use it. Maybe that's what we call "a clue?"

Kid says that’s about it.

Not the economy, just certain persons' wealth...

I saw someone on YouTube suggest that the current trend with AI is a way to "allow wealth to access skill whilst ensuring that skill can't access wealth".

I think that’s why they’ve always loved Tesla - it offers their ideal model for cars, as mobile data centres which constantly transmit our personal data and which we pay for, preferably via monthly subscription.

It is quite jarring seeing this directly alongside Sridhar’s ongoing lamentations for the loss of his field (pure math), once considered the pinnacle of human intellectual achievement and now just a playground for AI companies and their shiny toys.