archive: https://archive.is/cZZZC
Once physicians began using it, their performance dropped significantly whenever the system was unavailable. During the three-month period before the AI tool was introduced, the specialists found at least one adenoma during 28.4% of colonoscopies. During the three-month period after the tool was introduced, the adenoma detection rate for colonoscopies performed without AI assistance decreased to 22.4%.
Afterwards, all of the software engineers were asked to complete a quiz about what they had learnt from the task. The participants who had used an AI assistant did significantly worse on the quiz than those who hadn’t: the average score was 50% in the AI group versus 67% in the non-AI group. The AI-assisted participants did particularly poorly on questions that required them to diagnose errors in the code, which suggests that they had failed to learn the concepts behind the code that they had just produced.
I believe AI has become a tool that is both good and bad, just like the internet, but on a large scale, it has advanced exponentially. But tell me, whose responsibility is it if it's good or bad? It should be used as what it is: a tool to advance in all the areas where it's applied, whether it's health or future projects.
You're getting bad at the things you let the AI do which you need to test for without the AI.
Like you'll get bad at shifting gears in an automatic car, and then over time will perform worse to other drivers who've only been in manual cars.
On average we all ruined our horse riding skills, no doubt.
I'll be honest. I am inclined not to believe this, or at least the framing of it. In my personal work, especially work for which I'm already deeply familiar, I don't think that a 3 month window of AI use is going to cause me to perform significantly worse on things that I'd already been doing for years.
I agree with you for almost anything except one thing: boilerplate, not designing them; using them. I just can't be bothered to do it anymore. Not even copy & paste. But, I'm not letting AI win that one either because that would be inefficient. Instead I just make cli commands and if it is some recurring thing I crontab it. And thats it, problem solved. No more tedious repeating myself.
This was caused by LLM usage though - simply because the amount of boilerplate in interacting with LLMs went through the roof.
I would guess that in cases mentioned in the article it's a loss of vigilance more so than knowledge.
This is directionally the same as what I was trying to express here: #1509787
Unless ideologically motivated, who wants to do what they don't have to do? The option that an llm could save you the time or the trouble of figuring something out is like raising the groundwater level: all the voids may be filled, but it doesn't make a better foundation. (Possibly this analogy needed an llm to clean it up).
If we can think of our minds as muscles, I have no doubt that we will all see some atrophy in the next decade.
I would assume that these software engineers have attained a level of mastery and would be discerning users, so it is concerning that in spite of their judicious use, they still suffer from a decline in their cognitive abilities
after looking at the centralized medical curriculum and contemplating the thot process of the academons-indoctrinated physicians:
they really do not know shit about fuck, however they sound & feel confident in their nescience & ignorance;
i am sick of witnessing them butcher people & lie to their face (while doing the same even to themselves & their families);
the medical order-followers hurt way more people than the military order-followers, and it's all by [not very well informed] consent: contemplate!
How to correct grammatical errors without AI? What is your suggestion? Waste 3 hours searching for grammar and similar tricks is outdated.
scary times