it’s always interesting how often true innovation ( not to mention the best art and music ) come from people with free time.
by locking people into jobs until the reach the grave, we’re denying not only individual growth; but society’s.
6 Likes
30 years in engineering and business leadership.
I’m all for automation of processes through algorithmic optimization.
Calling that “AI” is a reach, though. The problem I have with real AI is loss of human control of the algorithm, if you will, as well as attributing “reasoning” to what is basically a black box containing a succession of arbitrary value judgements. Most people who I’ve seen use AI for business decisions would have done better to either do their homework themselves or ask an expert.
14 Likes
The comment I was laughing at reminded me of a company I used to work for. They brought in a consulting firm to find out, among other things, what factors most influenced employees to stay or leave the company. The fact that things like job satisfaction and having a challenging work environment and work-life balance were rated higher than things like pay was used as justification to not give people raises. Because obviously people weren’t motivated by their pay.
People working into their 70s and 80s is the same thing. They may find their work engaging and enjoyable, but that does NOT mean they wouldn’t retire in a heart beat if they felt they could afford to.
ETA: My uncle was an archivist. He worked for Harris County, Texas for many years. He loved his job. I mean he loved it. He used to joke all the time that he hoped his bosses never found out how much he loved his job because he was afraid they’d ask him to do it for free. He still retired 5 or 6 years ago at the age of 70 or 71, because he could afford to retire and he was tired of working.
10 Likes
Off the shoulder of Orion.
That’s small potatoes for government work.
4 Likes
YMMV. In my industry knowledge of government rules, regulations, and laws are essential. The LLMs I’ve seen so far hallucinate responses, are flat out wrong, or miss important information and are often unable to successfully complete queries. They’ve been no match for learning over 4000 pages of documents.
As for mundane, if only an LLM could prepare my TPS report for me, that would be great.
The schools, roads, bridges, and waterworks (wells, micro dams, irrigation, and local water distribution) projects I’ve financed or managed over the years were initially prepared on AutoCAD or similar program. The outputs I saw were in PDF, but any revisions were done in the program. The only time the architects and engineers I hired would work off a pdf or paper document is if it was to perform work on an existing structure from drawings filled in hard-copy form at the relevant government office. Unless it was a simple project, the firm would somehow replicate the plans in AutoCAD to do their work.
Dammit! Why won’t the exorcism stick?!
This sounds like a job for Microsoft Excel! (No need to drag an expensive LLM/AI system into it)
As an aside, I had an employee who would calculate loan appraisals in a table in word, manually updating each cell when a variable changed. She literally cried when I taught her how to use excel. This was also in Kyrgyzstan. Twenty years ago. And a person whose first experience with a computer was literally 2 years prior at a different job. She was a brilliant woman and wound up teaching me several useful “tricks” in excel, some I still use today.
And there are young people effing up well established processes because they insist on using fancy new tools. I’ve been labeled a Luddite because of my resistance to doing certain processes in MS Teams, even after I provide specific detail on what can go wrong with the process and why it is essential that this not happen. Then it happens, and everyone gets upset when I pull out the communications telling them it would happen and insist we go back to the tried and true method. So it happens again. The loss in revenue and productivity has been astounding, but we must use the shiny. (I am also one of the Teams power users at work, ironically.)
AI has done some impressive work. What I’ve read about it’s work in calculating protein folds is amazing, though I wonder how much is really “AI”, and how much are fancy programs or algorithms doing the work.
But the idea that AI is going to lead to increased efficiency and productivity hasn’t panned out in the industries and specific areas my work touches. Just like blockchain. And bitcoin. And social media. And a whole host of “innovations” that were supposed to revolutionize my field of work, yet somehow didn’t. Color me skeptical.
12 Likes
… if it’s a multidimensional matrix of so many value judgements that no human being could ever understand them, it must be good, right 
6 Likes
9 Likes
And you just know Lumbergh would run all the reports through AI to get a summary.
6 Likes
Yeah, because they are entirely responsible for the entirety of the economy and all the ills of neoliberalism…
And from what I’ve heard from many in the field is that it’s not much better than what’s available to the public.
It’s well known that tech journalism is in bed with silicon valley, so if they had some real successes there, they’d be covering it as proof that AI is not the BS that it seems to be…
Jobs WILL be cut. That’s what will happen. This story is as old as capitalism itself. It’s what the Luddites were up in arms about.
Hard is not impossible.
The problem hasn’t been “bad data”… it’s been LLM’s outputs. And the endless scrapping of copyrighted materials, with the output being… the same material, pretty much.
I’ve heard that’s the problem of the data, not the AI! The data is clearly failing the AI! /s
That’s the hype-cycle mentality… There is nothing wrong with new tools, but how about not embrace everything that silicon valley shits out without thinking about it or giving it some robust testing?
Still not convinced of that, sorry!
That seems relatively true across industries… I doubt anyone is gonna get a good history paper out of any of the AI programs laying around right now.
7 Likes
Really? Mostly it just churns out averaged shit. You must have the Executive version.
7 Likes
In medicine and biology, it has been impressively helpful in doing some brute force work and identifying signals that lead to diagnosis of such maladies as cancer. (Science Friday has quite a few stories about these successes.)
However….
It is unclear to me if this is some well crafted program or algorithm with crazy processing power behind it, or the sort of “AI” being hyped by Silicon Valley.
The funny thing is these advances are building on work and successes predating the current “AI” fad (as reported by Science Friday), so it could be a case of slapping a sexy new name on an existing system to get attention. 
Also, there is always a human expert at the end to evaluate the output. The machine is never in charge.
6 Likes
In molecular biology, “AI” is really making breakthroughs in throughput - if you define a rapid digital image capture and pattern recognition algorithm as AI. Some do, some don’t. Notably, when a company is marketing their new technology or looking for funding, it’s called AI. When the designers are talking with actual molecular biologists, not so much.
ETA: @MagicFox ninja’ed!
7 Likes
@DukeTrout - I’m gonna have to take your words for that… I hope that’s the case, but it seems to me that we’ll have to see how it comes out and if there are problems down the line.
That could be it?
But I do know for my field, it’s just… crap. It’s gonna be (already is) a plagiarism machine for desperate undergrads…
7 Likes
I saw recently that one program designed to identify cancer cells had an interesting quirk.
The images it recognised as cancerous tended to have been submitted by doctors and had measuring devices for scale.
Those recognised as non-cancerous tended to have been submitted by the general public, looking for diagnoses, so were not labelled or scaled.
In effect, the “AI” programmers had invented a program only really good for identifying rulers.
10 Likes
No, no! My experience has also been that it is crap. Grade B quality fertilizer, don’t put it in your roses if you want them to live.
@DukeTrout summed it up better: it seems more like companies and researches labeling an existing technology “AI” for marketing purposes.
8 Likes
Science Friday had a segment on that as well!
(And I’ll leave it at that, ‘coz I’m not here to defend AI.)
4 Likes
Funny, I thought you had articulated it better. Or at least more clearly stated that a lot of what is being hyped as AI is questionable at best.
4 Likes
To my ears, you sound like you have some fantastic bridges for sale.
Why, it’s almost like it’s a hustle.
7 Likes
I imagine if somebody’s current business process consists of a bunch of people bullshitting each other, then those people might easily be replaced by AI
and yet 
7 Likes
You mean to tell me we’re spending too much money on schools? infrastructure? and yeah healthcare is a money pit-- but you’re saying people are going bankrupt over medical bills because too many workers aren’t Excel super users? Woof!
5 Likes
