I dont think so. Admittedtly I live in a peculiar bubble (mix of scientists, artists and programmers) but nobody that I know personally talks about chatbots in neutral terms, let alone in favorable ones.
If we take chatgpt and the ones from the other brands as neutral entities we as a society have a lot of work and self-reflection to do.
Quoting a chatbot in an article that is not related to a chatbot felt offputting and weird. I had to stop reading, because it somehow undermined any interest i had in the article
difficulties in life and genetics are more of a reason to decelop a severe addiction. Quality is probably not so much of importance - people still smoke cigarettes and drink alcohol.
It feels like we forgot about the opiod epidemic, meth, crack..
Anyhow, I think this derailed enough. I just wanted to point out the weird protestant views of op.
>Humanity has discarded natural selection thanks to modern medicine. Gene mutations that would have meant someone didn't survive now can be treated.
There is still selection going on and it is difficult to argue that it is not natural. The pressures we are exposed to are just not consistent with some idealized natural state and thus seem "unnatural".
Be careful, "natural selection" is a specific descriptor that describes a selection process that is contrasted by "artificial selection". The second one comes up from time to time in human context; we call it "eugenics".
I'd wager it is the strongest. Most of the arguments always argue in quantities. So and so much of energy waster per hour, this many job losses and that growing trend of delusions supported by LLMs.
We've already seen it with climate change. Arguing with statistical factoids is too ethereal for anyone and leads to some kind of fatigue. There is no emotional difference between 100 mWh and 10000 mWh. But a "I had to bring my dog to the vet because chatGPT told me to give it chocolate" will convince anyone to deeply distrust technology.
FAANG and its acolytes deprived us from actual connection with our environment. Judging the argument as weak is sad proof of that.
So how about instead of using the slot-machine that chatGPT is, we go visit our grandma, hug our dad or just get a new houseplant :)
I checked the BlEU-Score and Perplexity of popular models and both have stagnated around 2021. As a disclaimer this was a cursory check and I didn't dive into the details of how individuals scores were evaluated.
>I feel the opposite, and pretty much every metric we have shows basically linear improvement of these models over time.
Wait, what kind of metric are you talking about? When I did my masters in 2023 SOTA models where trying to push the boundaries by minuscule amounts. And sometimes blatantly changing the way they measure "success" to beat the previous SOTA
How "subtle and insidious" is it really?
I'd say it is shifting the blame of personal responsability to a website. Me and some of my friends use(d) 4chan and we never fell into the pipeline. To the contrary there is a strong left-wing camarederie. And I'd wager that we recognize subtle right-wing views more easily. One doesn't learn about these views by looking at a twitter screenshot but by engaging them.
We should stop treating right-wing ideology as a mind-parasite. And if we do it anyways, we should accept that some people want to get "infected".
If we take chatgpt and the ones from the other brands as neutral entities we as a society have a lot of work and self-reflection to do.