The road to hell (and stupidity) is paved with AI

For years, I've been reading about how our abundance of technology is changing the way we think - and not in good ways. Cursive writing, using maps, reading books and pretty much any other form longer than the average tweet, making to-do lists on yellow pads. All of these skills aren't just lost arts. They're important, as they help us with cognitive development, memory retention, critical thinking, analytical ability...

And this was even before AI use cases started soaring. 

With AI making inroads, the law of unintended consequences is clearly being enforced. It's not just that folks relying on technology to do just about everything for them involved in thinking and doing won't develop key capabilities. It's that using AI will cause their existing skills to erode. 

In one study, doctors who'd gotten used to relying on AI to identify pre-cancerous growths during colonoscopies saw a decrease in their own ability to do so. 

This deskilling is worrisome to a lot of folks:

That possibility is a growing concern for medical specialists, computer scientists and other workers. Seventy per cent of nurses and 77% of physicians, for example, are worried about losing their skills because of over-reliance on AI systems, according to a survey of US health-care workers published earlier this month1.

Their fear might be justified. Evidence suggests that AI-driven ‘deskilling’ is starting to happen in medicine, computer science and other fields. Researchers are now discussing how to preserve important human expertise in the age of AI. (Source: Nature)

On the one hand, what of it? If from here on out there'll always be AI around to do the growth spotting, what difference does in make if a doctor's not as good? 

Good question, but I'd prefer my physicians to be able to, if not heal themselves, then heal me without all the heavy lifting being done by an AI. What if we're stranded on a desert island? Although on a desert island, I'm not likely to undergo a colonoscopy, I'd still want a doctor who knows how to do things the old-fashioned way. 

I have a friend who's a retired physician. She takes her blood pressure using her old hand-pump BP device, the one with the squeeze bulb. She trusts its reliability more than she trusts the swanky new digital machines.

Overall, the growing reliance on AI is not an unalloyed good for the wonderful world of health:

The study authors say that continuous exposure to such [AI] tools can cause clinicians to become “less motivated, less focused, and less responsible when making cognitive decisions without AI assistance”.

Less responsible when making cognitive decisions without AI assistance? That doesn't sound good.

Another example from my friend the doctor: She recently had a healthscare based on a misreading of a result. (Not sure if the situation involved human or machine error.) There was a slight problem, but not the major one initially reported, which had her writing her obituary in her head. Anyway, my friend's PCP was not particularly well-versed in what was happening to my friend and why. This is perfectly understandable, as the situation was a bit off the beaten path and no PCP can be expected to know everything.

The PCP went to the google and, we're guessing, AI'd her way into ordering a bunch of diagnostic tests for my friend, most of which my friend knew had nothing to do with the condition her condition was in. She played a bit of a waiting game, pushed back a bit, and ended up only with the tests she needed.

The situation is not, of course, perfectly analogous to the AI colonoscopy one, but it does demonstrate how a doctor might rely on what the technology is guiding them to do because it's quicker and easier than trying to figure out what actually makes sense in a given situation.

Then there's the impact of AI on coders.

By now, we're well aware of the looming coding takeover by AI. Folks used to look to computer science, to IT, to software development, for solid, lucrative careers. Now students are advised against studying much beyond how to work with AI.

Anthropic (an AI research company) recognizes the inherent problem at play:

AI creates a potential tension: as coding grows more automated and speeds up work, humans will still need the skills to catch errors, guide output, and ultimately provide oversight for AI deployed in high-stakes environments. (Source: Anthropic)

Anthropic decided to see how AI would help and/or hinder codes. So they set out to test the assumption about whether there were tradeoffs when it comes to using AI to assist - note, not replace - coders. 

Study participants were sorted into two groups, one using AI and the other "manually" coding, and were tested afterwards to see how rapidly they picked up a new coding skill 

The AI group was faster, but in the tests adminstered after the trial, they found:

...a significant differences in test scores the AI group averaged 50% on the quiz, compared to 67% in the hand-coding group—or the equivalent of nearly two letter grades (Cohen's d=0.738, p=0.01). The largest gap in scores between the two groups was on debugging questions, suggesting that the ability to understand when code is incorrect and why it fails may be a particular area of concern if AI impedes coding development.

The good news, I guess, is that AI is not quite ready and able to replace all of us humans. The bad news is that while it's making its way to full readiness and human-ish capability, it's making us dumber.

Swell!

-------------------------------------------------------------------------------------

Image Source: Parker's Ponderings