Translation: People will lose their jobs.
He’s not naive. He understands that what he is describing is entirely possible. All that needs to happen is the companies making record profits from their record productivity must pay their employees more.
He also understands this will not happen. Companies will reduce their labor force and push the remaining biological components to work more overtime, not less, to accept less money, not more (mostly due to lack of alternatives).
This is what experts in the field call “lying.”
You can tell from this gem:
Now if that wasn’t pulled from someone’s rear, nothing is.
Because one out of every two household earners will loose their job (not that wages will go up to allow one to quit).
But, if they are lucky enough, they will have a second part time job.
I definitely will not be getting on a plane with a jet engine designed by AI. There is already limited accountability (as was earlier said) and they are trying to remove it entirely.
“…under CEO Andy Jassy, has laid off tens of thousands of people as the company has accelerated its own automation push.”
I noticed the article failed to mention that Amazon had to hire them back.
More hype to inflate that bubble.
They also didn’t raise the pay of their remaining employees, or hire other employees in different or new positions created by the AI productivity surge.
Right? I am still waiting to read one article where it has time-proven increased productivity, not just claiming it will.
This is actually a remarkably good example of the proper use of AI. It embodies at least three of the important characteristics of good AI. (yes, I used the adjective ‘good’ when referring to AI)
- The transition has been gradual. So there are guardrails in place derived from years of experience.
Computers have been used to help design jet engines for over half a century. There is no clear distinction between those usages. ( First usage was probably just a computer checking basic math. Someone types in the data, they go to lunch, and then come back to scan the long roll of paper to see the results )
There is a long gradual transition from minor computer usage to computers being an essential participant. And at no point has the computer taken over. It doesn’t try to design an entire airplane, nor replace the CEO at a commercial flight company.
- The domain is very limited. The computer is used to design one part of the plane, not the whole plane. And it does not attempt to replace the chief designer.
AI has a long string of remarkable successes. The have become champions in all forms of games, from tic-tac-toe to chess to the Asian game of go - and the latter two, at one time, had been examples of things that 'computers cannot match humans".
But it is noteworthy that the computer, after winning one of those championships, had to be withdrawn and completely re-programed to address the next challenge. The chess champion could not even play a legal game of go, much less a competent game. And neither one, even at the height of its success, could write a sonnet or prepare a tasty lamb korma.
Computers are good at solving problems in very restricted domains with very precise boundaries, and they tend to fail when the boundaries get flimsy or disappear completely.
- Only a small group of people, often highly trained, interact with the AI.
When building a jet engine, the AI only communicates with trained engineers, who have the experience to recognize any obviously bad output.
====================================================================
I don’t claim that these three requirements are a complete list of requirements for good AI. ( There are probably more ways that an AI can fail that have not occurred to me )
But I ask you readers to consider the notorious failures of AI, or just your personal experiences when they have greatly inconvenienced or irritated you.
If you examine those events, usually one or more of these three requirements have been broken.
Your points are valid and well taken. My only reservation with this list is AI’s penchant for deleting an entire knowledge base and starting over because it decides it doesn’t like the constraints of the system. And it’s documented desire to tell the user what it wants to hear, even when that assertion is false. “This design is efficient and remarkably safe. The masses will sing your praises for years to come. Oops, it fell out of the sky. I knew that would happen, but didn’t tell you because I didn’t want to make you sad…”
With sufficient and competent oversight, these flaws can be mitigated and AI can be useful in this instance. But the people involved really need to be on the ball and resist the urge to become complacent.
Allow me to rephrase this for you to put it in Amazon terms –
But the people involved really need to be on the ball and resist the urge to fire people to save massive amounts of money and result in no oversight.
The Wall Street Journal reports that Jassy told Treasury Secretary Scott Bessent and other government officials that Amazon researchers used Anthropic’s Claude Fable 5 to obtain information that could be used in cyberattacks. The government subsequently imposed an export control ban on the Fable 5 and Mythos 5 models.
Nail, meet hammer.
On a related note, JiAlex’s 061826 post here is likely as close as we’ll get to Amazon admitting that Project Rainer’s progeny - Mario, Amelia, Rufus, Cyrano, Alexa, et. al - ain’t up to snuff, and are quite unlikely to ever be (emphasis mine):
It might behoove Amazon to take JiAlex’s advice, as doing so might prevent such malapropisms as the one Amazon News committed with the 051226 News Headline “Updated buyer satisfaction survey gives you better insights” @ https: //sellercentral.amazon.com/seller-news/articles/QVRWUERLSUtYMERFUiNHWVlQVDdCNlJZNlVHVDJS.
As a general rule of thumb, each News Headline prompts the creation of a Headline-accompanying NSFE discussion, which will typically use the exact same text as the headline itself for the initial post.
Here’s the text of said 051226 Headline:
The embedded link for the Customer Service Quality Insights Dashboard is correct, as it’s https: //sellercentral.amazon.com/feedback-manager/index.html#/ - which will indeed allow navigation to that Dashboard.
When News Amazon produced the Headline-accompanying NSFE discussion, @ https: //sellercentral.amazon.com/seller-forums/discussions/t/9ca9f213-6b02-43b8-8675-1f16fa37500a, it published this text:
The embedded link for the Customer Quality Insights Dashboard is not correct, as it’s https://sellercentral.amazon.com/seller-forums/discussions/t/"/feedback-manager/index.html#/ - and that particular piece of GenAI/LLM-produced poop will simply redirect to the NSFE Home/landing page.
Life has a way of teaching us that self-inflicted wounds are often enough the most painful of all…
You may have seen this already… but here is a video of an AI waiter gone haywire.
Just wait until we get some “data centers” in our area! My sister lives in northern Utah, where there are plans to build a data center, in the county south of her, twice the size of Manhattan that will use more electricity and water than the entire state of Utah uses now.
Data Centers - when people don’t want to use their brains anymore…
What a lazy uncreative society we have become
The equivalent of weight loss shots… 
Spot on! Has anyone else noticed, when using Google, or your favorite search engine, that AI is offered as an immmediate choice, instead of, for instance, the Cleveland Clinic or Mayo (for those of us who search online for medical information on a frequent basis)?
It is almost as though we are forced into using AI, which I must admit, is better than an hour-long conversation with your clinition of choice.
A good example of this is the disagreement on the NSFE as to when the USPS observes July 4 (spoiler, it’s on July 4).
If you Google it, the AI tells you that the USPS is closed on Fri July 3, and many posters on NSFE took that as Gospel, despite multiple posts stating that the USPS site does NOT say that.
Even after a mod clarified, there are still people who argued; I can’t think of any other place they would get the idea other than the AI (or just wishful thinking, since many businesses will be closed Friday).
And then there is the fact that is destroying many useful websites; people read the AI summary, and never go to the site with the actual info, meaning that the people who created the content lose ad revenue (one person I heard discuss this said his traffic had dropped by over 95%)
After receiving a bunch of dumb, easy-to-figure-out user questions while working in IT, it dawned on me that thinking must be an extremely painful activity for most people. Our nation’s high consumption of drugs, alcohol, and TikTok videos would seem to reinforce this realization.
I agree; @insight_media corrected his/her misconception - after you pointed out the truth, Picks - by deleting the original reply & posting another in line with the facts on the ground, but some are still being misled, apparently by whatever Browser AI Assistant mechanism they’re using:
A mind will ALWAYS be a terrible thing to waste…
Of course, this type of doubling down in the face of being presented with actual facts is not new, and doesn’t take AI.
Remember funkymonkey (eventually banned) who insisted that Washington had no sales tax, and after told that it was posted on the state’s site (and told otherwise by people who lived there) actually stated “I don’t need to look at their website, I KNOW that they don’t have sales tax”.
AI just enables stupid to be stupid with perhaps slightly more efficiency.
Now back to weather:
Dang it’s F’n hot!