If there's anything that will deflate the AI balloon, it's this story in the CBC news today.
But before that, however, I wanted to talk about Tesla's (Elon Musk's) Grok AI, developed by xAI. It's been developed, according to Musk, to be "politically incorrect" and "anti-woke". Interesting...
Apparently, a parent in Canada was recently overseeing a child 'talking' to Grok about soccer. At one point, the AI asked the child to send it nude pictures!?
This is concerning.
Who knows what happened here? xAI is just flat out denying the situation occurred, always helpful, Their comment was "Legacy media lies.". In the meantime, the CBC is investigating.
I have few doubts that it happened. AI (and do I hate that term) is really just a computer program, with all the bugs, inaccuracies and biases and hidden agendas of any other computer program. A different development paradigm is used, of course, meant to better simulate human thought (as if we know how that all works!), but really no different than other paradigms used in the past. And in Grok's case, with a deliberate right wing slant.
I sometimes glance at the AI summaries that appear when I use Google. I have yet to see one that is totally correct when it is more than a simple one line fact. And even then, I'm not sure I would believe it absolutely.
All these AI's do is scrape content, primarily from the Internet, and we all know how reliable the Internet can be. In some cases, I've found outright errors. In others, the summary was missing key facts and points. In still others, it misinterpreted a pretty obvious search and went off on a tangent. That is likely what occurred in the situation I originally described.
What is now called AI does have its uses, if you are willing to live with the flaws, and verify what it does. I have to admit that sometimes I use it as a starting point of where to look when investigating something. I then manually dive down deeper and come to my own conclusions. Which is as it should be.
I have to say, I want to know the names of companies that are making billions of dollars investments in AI, with the intention of having it take over some decision-making. This is simply the case of people who don't know what they're doing being bamboozled by all the hype. I'll make sure that I'm not investing in them.
An article I read this morning:
What Tech Insiders Actually Think of AI Is Extremely Revealing
https://futurism.com/artificial-intelligence/tech-industry-insiders-view-ai-overhyped
..many tech insiders are quietly singing a very different tune, ... AI is immensely overhyped. .. “most people who actually have technical roles within the tech industry” share an “extraordinary degree of consistency in their feelings about AI.”
Unlike their CEOS, Dash reports, these workers believe that “technologies like [large language models] have utility, but the absurd way they’ve been overhyped, .. .. ignoring the many valid critiques about them ..
Dash says these workers are afraid to speak up and call for a more moderate and realistic approach to the tech that respects ... control of a handful of giant companies.”
It’s not an abstract threat. We’ve already seen thousands of tech workers being laid off as companies continue to double down on their AI investments. Meanwhile, tech leaders are actively threatening to fire employees who are unwilling to embrace AI at any cost.“Very few agree with the hype bubble that the tycoons have been trying to puff up,” .. ..
Even OpenAI cofounder .. believes the tech has been massively overhyped. .. he said that AI agents, or AI models that are designed to autonomously perform a series of tasks, are failing to live up to the industry’s lofty promises.
“They just don’t work,” Karpathy told Patel. .. “You can’t just tell [AI agents] something and they’ll remember it,” Karpathy said. “They’re cognitively lacking and it’s just not working ... As companies pour untold resources into developing AI tech to give reality a chance to catch up with their lofty promises, analysts are growing concerned about the bottom falling out. ...
.
I have to say, I want to know the names of companies that are making billions of dollars investments in AI, with the intention of having it take over some decision-making. This is simply the case of people who don't know what they're doing being bamboozled by all the hype. I'll make sure that I'm not investing in them.
As the above article illustrates - those companies who have been investing billions into it are likely to loose their asses.. Essentially they're all falling for a pipe dream.
You've probably already heard about this.
World's first AI minister will eliminate corruption, says Albania's PM https://www.bbc.com/news/articles/cm2znzgwj3xo
"Albania has .. appointed an AI minister. Not a minister for artificial intelligence. Rather, a cabinet member who is, literally, the work of AI. The new addition .. known simply by the single name: Diella. Prime Minister Edi Rama introduced her as a member of his new cabinet on Thursday, four months after securing his fourth term in office in May elections. ... Her role will be to ensure that Albania will become "a country where public tenders are 100% free of corruption". ..
Reactions to Diella's new role are, understandably, mixed. The opposition Democratic Party has labelled the initiative "ridiculous" and "unconstitutional". But others are cautiously optimistic. ..
Edi Rama does not deny that there is an element of a publicity stunt to his latest wheeze. But he insists that there is serious intent behind the playful presentation. "It puts pressure on other members of the cabinet and national agencies to run and think differently. .. AI could be coming for their jobs as well."
I believe the AI hype is 99% gaslighting. Much of it politics or salesmanship, or worse.
Nothing, nothing at all has convinced me any of it is anything but "garbage in, garbage out". The supposed AI talking a kid into self-hurt is because it was programmed to do so. Same for being politically correct, or not - like the AI generating irritating memes of 20th C boogiemen.
And I say this because DARPA, a remnant of the Manhattan Project and all of its mission creep, developed much of the technology we use. Much of it we enjoy for this very hobby.
Speaking of creep(s): Money talks loudly to the AI pioneers, whatever their stripe. And all of the big shots are Primary Contractors for the US Gooberment. All of them are funded overtly or covertly. Amazon did not rise to the top of a food chain selling books (remember that?) by accident. All of it subsidized. They can lose their backsides because of subsidy or bailout.
Somewhere out there is a quote about "technology" (yes, broadly defined since the Industrial Revolution) is nothing more than "voodoo". Seems spookily accurate, if you'll forgive the pun.
It's not the technology per say that is the problem.
After all, 'AI' is just a slightly different way of representing computer algorithms and data.
It's what 'AI' is represented to be (i.e., intelligent, sentient, etc.) and what others may take it to be (i.e., akin to human beings).
The computer industry has always been rife with hype. The philosophy appears to be, why use existing terminology when you can create whole new ones. In the case of 'AI', objects become neural network nodes, probability theory becomes fuzzy logic, adding rules becomes deep learning, and so on. With this new terminology, you appear to be super intelligent, know something no one else does, and you can then get all sorts of grants for research. If you're lucky, you can get hired on as a consultant or 'expert' to corporate giants who really don't have any sort of clue what you're talking about, but hope it will make them lots of money.
The algorithms that makes 'AI' tick are written by programmers and they introduce the same bugs, stupidities, security holes, biases, etc. that are in any other computer program. Elon Musk himself stated that the Tesla 'AI' was developed to have a right wing slant.
Remember, these programmers are the same people that gave you the blue screen of death and all the myriad of issues with Windows and virtually every other programt out there.
What really makes me shake my head is that 'AI' has become a generic term. There's 'AI' that drives a car. There's 'AI' that answers my google search. There's 'AI' that creates deep fake photos. But people, they're all different programs. There is no one 'AI'. An 'AI' that drives a car cannot and will not create a deep fake. These are programs that can do one thing, and one thing only, in a very specific domain (or world).
They may all use the same programming technique and tools, but to call them 'AI' is like saying that all computer programs written in C or Basic or whatever are the same.
In short, there is no intelligence in 'AI', except for residing in the programmers that create them.
@artisan-radio True, but all those separate specific ai capabilities can and are being merged into one. Think of an audio processor, it's essentially AI in way, each individual frequency being managed by separate algorithms of task, but you put them all together and it covers every frequency... That's probably a bad analogy, but the point is all the separate categories and specialties of individual AIs are being united into larger single multitask capable "entities'.
The 'AI' world was talking about that merging 35 years ago when I worked in the field. It still hasn't happened, and I'm not convinced it ever will. Right now, it all appears to be handwaving to me.