AWS CEO: It's funny when people ask me if AI is overhyped

Stefan Weitz, CEO and co-founder of the Human[X] conference, welcomed attendees to the AI-focused bitshow in San Francisco with the promise that they would receive no certainty and no playbook. "You are moving at speed that you didn't choose through conditions you can't see," he said, echoing perhaps unintentionally Rod …

[Anonymous Coward]

Define efficiency

Crashing & wasting resources faster does not count.

[Blackjack]

Re: Define efficiency

https://www.theregister.com/2026/03/10/amazon_ai_coding_outages/

AI you say?

[I ain't Spartacus]

Re: Define efficiency

Define efficiency

Crashing & wasting resources faster does not count.

Efficiency is the speed at which a business executive can turn VC contributions (and/or sales income) into bonuses for themselves - plus the percentage of said incoming cash that they can capture.

[that one in the corner]

$4,000 price of admission

This is just squeezing more cash out of the gullible before it all pops.

Then again, it's going to be cheaper than attending the football World Cup and about as useful in the grand scheme of things, so what the heck.

[Doctor Syntax]

Re: $4,000 price of admission

Came here to say much the same thing. They've obviously got the measure of the depth of their audience's pockets.

[Anonymous Coward]

Re: $4,000 price of admission

I see potential for 10x growth. I'm sure true believers will be happy to pay that much. It'll just weed out the riff-raff skeptics.

[ecofeco]

4000?!

All your base are belong to us!

[Groo The Wanderer - A Canuck]

Sounds exactly like the same odious bullshit the other AI companies are spewing to me, regardless of the misleading summary and title.

[Richard 12]

WTAF?

The electrification of industry, he said, didn't show up as productivity until decades after factories plugged in.

That's so stunningly false that the audience were probably shocked into silence. Anyone able to say those words has drunk so much Kool-aid that they're in a diabetic coma and need immediate medical help!

Electrification improved productivity at every single step. Eliminating the common line shaft alone was a massive productivity improvement - a 30% increase in output is widely attributed to have occurred almost overnight.

This massive efficiency jump lead to huge reductions in prices, and arguably created the mass consumer market for the first time.

It was the exact opposite of this AI boondoggle -nobody connected their factory until the RoI was clear.

[The Travelling Dangleberries]

Re: WTAF?

I am really interested to find out how AI is going to make efficiency improvements to the firewood stacking process that takes place every spring here at The Travelling Dangleberries Towers.

[Steve Davies 3]

Re: efficiency improvements to the firewood stacking process

All it could do would be to optimise the stacking but only after every log had been scanned first. No time saved.

I've spent most of the day stacking the wood for burning next winter so thanks for the great post.

[Irongut]

Re: WTAF?

I'm sure the guy from AWS knows nothing about the electrification of industry or indeed industry.

He works for an overgrown shop that runs its own IT and rents it out. He's never produced anything or worked in industry in his life.

[I ain't Spartacus]

Re: WTAF?

Richard 12,

Not to mention the massive societal change that the electrification of lighting brought on. Which literally changed everything from education to working environments to uses of free time spent at home. Although, I guess you could argue that gas light, where you could get it, was already doing some of that.

Someone made a brilliant website on the cost of light compared the wages from an hour worked, showing how much it dropped over the years, with different technologies, and the changes that made in society. From an era when you were having to read/work by fire and candlelight, which meant that not much got done after sunset, and you just had to go to bed, because the alternative was a vast expense or sitting in the dark.

It also takes a uniquely stupid person to think that people will expensively retool an already working factory for no gain.

Even in the 19th Century, we had the equivalent of modern venture capital. Just read up on the railway financing scams and scandals. We changed the law on the ways publicly traded companies could be run, so that a bunch of these scams became hard, to impossible. I guess that's why so much VC money is poured into private companies - and they're only sold to the market when they can stand the public scrutiny of an IPO.

[werdsmith]

Re: WTAF?

Electrification improved productivity at every single step. Eliminating the common line shaft alone was a massive productivity improvement - a 30% increase in output is widely attributed to have occurred almost overnight.

Electric didn’t replace the common line in the first instance, industry didn’t just throw away all of its equipment overnight. Instead the line was powered by electricity and the flappy belts remained. It took longer to replace the equipment with new stuff that contained the motors. I’ve seen a sawmill still using common line in the 1970s.

[Snake]

Re: WTAF?

It's the fact that electric systems, with only few exceptions, *worked* once they were implemented out in the field. Failures were a rare occurrence.

"AI', not so much. They even create a new euphemism, "hallucinations", to delude and dilute the truth of the abject failures. When an 'AI' does something wrong it's just another day at the coalface of computing; when the electrical grid had/has a hiccup people absolutely paid attention, and most likely had to compensate for this major disturbance in their lives.

The comparison isn't even on the same continental shelf of connection.

[Random as if !]

Talks Pure Shite

To give it the Scottish angle, and good luck reviewing what you don't understand , mind you I'm retired now, but can still smell shite.

[Guy de Loimbard]

Re: Talks Pure Shite

Fine words!

I can hear the Scots tone as I read it and laughed as I think it aptly describes this whole movement.

More of the usual "you're not using the AI right", "it's you skeptics that are making it not work" and " Give us more money".

[that one in the corner]

70 percent of audience members had raised their hands.

We have an audience who paid a stupid amount to attend a talk fest (the sort of thing that could, on the face of it, be equally functional if run as a free webinar).

No, you at An Event. The added value of attending something like that is to be SEEN to be attending, by the other attendees and, if you are lucky, by the bloke on stage. Even better, by the cameras! You are trying to be part of the In Crowd.

One thing you are NOT going to be doing is be caught booing - or be spotted not putting your hand up when asked if you had a positive RoI: "Did you see Fred from NearlyMegaCorp had his hand down? Loooosser! Should we sell his stock?"

[The Man Who Fell To Earth]

Re: 70 percent of audience members had raised their hands.

I'd have thought the added value of getting one's employer to shell out $4k for attending is the employer paying for ones networking to jump to one's next job. AI musical chairs is an industry unto its own. Or so I heard from a friend.

[mevets]

I love the chutzpah...

Don't worry about the lack of immediate returns; to get that quick of a turn around, you have to launch an unsuccessful war.

You just need the faith to destroy your existing infra structure, replace it with ours, and don't worry, you will be fine.

[Anonymous Coward]

Hype, pump, dump and get rich

That's what the AI bubble is all about.

I hope the bubble bursts and hurts AWS badly. Can't happen to a nicer (sic) company.

[Jamie Jones]

Who owns the code?

I still haven't had a clear answer to who really owns the code "created" by AI. If it's all just clever pattern matching, then someone somewhere will be the original author, and what happens if that code is under GPL, or a proprietary license, or isn't even public, but just leaked data that has been snaffed?

it's a very brave organization that puts their copyright on code some AI has generated for them.

[find users who cut cat tail]

> Maybe the best skill that you can learn is learn to learn…

Sound advice. Logically then, preventing people from really learning anything is where AI excels.

[[e-mail]]

You don't go to an AI conference unless you are think AI is going somewhere.

[Groo The Wanderer - A Canuck]

Sure you do - you go to see which "big names" are attending and are likely targets for your "AI business" to cold-call. Not because you believe in your own marketing, but just to line your own pockets...

Mind you, greed or cruelty may also simply be an absolutely grotesque level of blind incompetence; the three are often indistinguishable in their end results.

[mevets]

Gotta call you on that one.

There are lots of freak shows that you can fully enjoy without having the slightest intellectual investment.

In a true circus, like a MAGA Rally, the audience themselves is a source of pure delight, never mind the tiny-parted people lined up to speak.

Take in a BigFoot Conference, you are sure to be cornered by some doped up enthusiast bubbling over about the incontrovertible evidence found in the BigFoot poos.

Share some pizza at a Flat Earth meeting; you can gorge on a breathtaking web of intricate lies and flawed implications that would make an AI-influencer blush. Really tear into the one about how Antarctica is the crust of the global pizza.

Bunch of truckers getting together to cheer on covid... okay that was just annoying; but the sheer inanity was something.

People are bound to get up to all sorts of stupidity. Be a good sport, and help them along with it....

[KimJongDeux]

Facts on the ground

I'm a retired consumer of industry's bounty. In the last year I have not received, from any AI serf, even one reliable / usable answer to a practical real-world question. Plenty of responses agreeing that the previous answer was, in fact, incorrect. My image editing software subscription has gone through the roof "because of AI" tools which are hopeless. I have two Anthropics image editing tools which are the very latest but very feeble.

I'm sure I'll one day acknowledge these visionaries as gods. But right now they're as useful as 1988 Lotus 123.