Simple Known Dangers with AI, and Possible Future Dangers Too.

[Jason]

The process of coding by non-humans is complex but interesting.If you wish to peek into the near future, or our attempts to predict the future (always iffy), you could read this:

https://sdh.global/blog/ai-ml/will-ai-replace-software-engineers-heres-what-the-data-really-shows/

Thanks for sharing. Interesting article. I agree that predictions are usually iffy. And there are some assertions at the beginning of the article that aren't evidentially strong:

Nearly 30% of 550 surveyed software developers already believe their development work will be replaced by artificial intelligence within the foreseeable future.

That's a pretty small sample. And it's basically a poll. Who are these developers? What kind of code do they develop? Is AI in the workplace? What do they know about AI that they didn't glean from mainstream media? So many questions.

Current job market indicators reflect this emerging reality. IT sector unemployment spiked from 3.9% to 5.7% in a single month, surpassing the US national average of 4% as of January 2024.

One month? Really? How does that compare to unemployment in other sectors? The US is experiencing an economic slowdown. The IT sector includes much more than software developers. Is the software development sub-category experiencing similar numbers?

A comprehensive survey of 9,000 software engineers revealed that 90% consider job hunting significantly more challenging now compared to 2020.

This could mean something, but it could still be the economic slowdown. Many of the largest companies are shedding employees who used to do simple tasks or using AI as the first line of tech support.

Much of the article confirms what I said, particularly that AI-produced code has a large error rate (30%). Developers will have to oversee the code that AI produces. What will likely happen, and is happening, is that developers will use AI to assist in their work. The author hit that on the nose. Maybe in time, AI will replace developers, which will be sad since it's such a well-paying job.

As you suggest, predictions are iffy. Let's lay down our bets, and if we're still around, then the winner can collect. 

Thanks for sharing, Buster. It was an interesting article.

[Jason]

Today's Examiner had a  relevant article.

It makes me think I made a good choice in pursuing my degree in Health Science rather than Computer Science. There seems to be a glut of software developers. And employers using AI to screen resumes and cover letters is disquieting. It must be hard for your resume to be looked at if you don't know the exact keywords to get past the AI (there are some, apparently).

[buster]

Jason once commented on the error of calling AI intelligent. It's stupidity has shown up quite a bit on my information searches lately, and these are the sorts of things I don't remember being wrong before, at least so often. And I always try to  avoid any click that will include AI in anything I do on a computer. (Human to the bitter end.)

Example one on Google: I asked the simple question 'When will such and such a program release series 3?' The surprising answer was that this program would release series two in August, which helped me not at all because I'd watched series two in the spring already. A notice said the answer was generated by AI.

Example two, also on Google: There are two shows released a month apart in 2023 called The Diplomat. My searches kept going to the US Netfix series. So I typed "reviews for the diplomat BBC British series". That seems pretty clear to me, but it insisted on taking me to reviews of the US production.

I can't remember Google being that stupid. Maybe there is a switch that insists on using AI to formulate answers rather than some of the older algorithms that took you to the places that matched the limited clues given.

But at least with AI we don't have to wait as long for errors as we do with humans. That's one good thing.

[buster]

I was so impressed by the post that I did some searches and found this:

https://www.tomshardware.com/tech-industry/artificial-intelligence/cringe-worth-google-ai-overviews

[buster]

This solution seems to work but not always. I want to get rid of Googles starting with an AI Summary, where information can be gleaned by AI from unreliable sources.

So at the end of my search entry I put (space) -noai

General this jumps right past the summary into several pages found by the search engine.

[Jason]

Interesting article. Giving training material across the web as being equally authoritative is a really brain-dead idea, especially with the most dangerous of bad information, in health. I would say out of the webpages out there, probably 90% of the health-related sites are misinformation and often, disinformation.

As far as search engines go, I don't use Google unless I can't find what I'm looking for using a different engine. I figure they already have enough of my personal data. They also have singularly destroyed news media and given nothing back (because of stealing the advertising market). However, I will give them kudos for agreeing to pay an organization of news media in Canada, so maybe it won't all die. Facebook still refuses.

[buster]

This is mostly a rant against the loss of a far simpler world. This sequence of events started a week ago today, on the Friday, when Marilyn was going to do some shopping, and our old car battery decided it wanted to retire. Not only did it refuse to start the car, it was so weak it wouldn't even light up the dashboard. In fact the key had to be inserted manually to open the door. (The battery subsequently froze.)

So I called my car repair shop and found I could get this fixed on Monday if I could get the van towed there Monday morning. Apparently this would be easy.

So I called my trusty CAA membership number and found it was so busy because of the cold and snow that I would have to talk to a theoretically intelligent machine agent.

So the machine had to understand two pieces of information. Pick up  the car at my address early Monday morning. And then deliver the van to my repair shop. Any student I ever taught over my 33 year career would have understood these two simple things immediately.

You would burst into tears if I expanded our conversation, which for some unknown reason involved delivering the van to Younge Street in Toronto.

Fortunately I have a neighbor two houses down who works for CAA,  and I visited her and she was able to do some old fashioned connecting to real people and arranging the simple task. She said they were totally overloaded by this year's weather and had run out of humans to answer phones.

Jumping into AI, even choosing calculated response procedures, should be handled carefully. Human brains are a great and valued resource to depend upon.

I like people and their brains.

[ssfc72]

How old is your vehicle and how old is was the battery, Buster?

Maybe you can get one of the new Chinese EV cars, that are now not being tarrifed at 100% :-)

[buster]

I think this electric car might not pull the trailer very easily or very far.

The battery is 7 or 8 years old. It's a warrantee replacement of the original during the first year or so. Van's approaching 9 years old.

According to the always truthful Internet, modern car batteries average 3-5 years, which is disgraceful. That's built in obsolescence. So mine did well.

[ssfc72]

Yes, that battery did well to last that long.

The battery on my then new 2016 Subaru Outback only lasted 4 years, 1 year after the car's 3 year warranty, ran out.
Subaru did eventually come out for repayment of the battery due to the issue with premature failure of these batteries. :-)

The Canadian Tire car battery that I bought for the car is still working well at after 6 years and the Outback sits outside on my driveway all year long and still starts up at -20 C in the mornings.