bence pfp
bence
@bence
so with both the cmu and mit studies pointing to specific behaviours with ai tools reducing cognition and critical thinking, where does that leave us with daily use? first, let’s be clear what these studies concluded. using ai as a ghostwriter reduces every possible important metric for cognition and output. you’re less creative, remember less, and over time become less sharp. while using ai to challenge what you’ve already made, and build on it, actually improves these metrics baseline, i think what's being revealed is you need to be at the extremes, with either: 1. writing fully without ai tools, using it only to debate your line of thinking as an intellectual partner 2. automating your writing fully, with zero input from you, apart from historical data for training
1 reply
0 recast
8 reactions

androidsixteen pfp
androidsixteen
@androidsixteen.eth
Different use cases 1. Is a tool to enhance creativity work via research and inspiration 2. Is a tool to reduce overhead of boilerplate Both of these tools already exist, I think LLMs just turbocharge them Problem is, with a narrative around AGI / “no more jobs” meant to psyop people into overestimating the tool, the general population doesn’t know how to discern 1 from 2
1 reply
0 recast
1 reaction

bence pfp
bence
@bence
like this categorisation, you can almost work backwards from it to reframe the studies: research looked at ai creative work, and found that bc genpop can’t distinguish 1 and 2, in worse cases, ended up treating everything as boilerplate work
1 reply
0 recast
1 reaction

androidsixteen pfp
androidsixteen
@androidsixteen.eth
🎯 Overfitting bc the tool peddlers have a perverse incentive to make you believe it can do everything
1 reply
0 recast
1 reaction

Chris Carlson pfp
Chris Carlson
@chrislarsc.eth
Yes!!! Include some training sessions on how to use for 1 and 2 prior to the study taking place and get wildly different results, guaranteed
1 reply
0 recast
0 reaction

bence pfp
bence
@bence
what worries me is this should rly be obvious if previous market trends are anything to go by, when presented w the option of convenience at no greater cost, it will win i have 0 faith in avg user to delineate between these use cases and make healthy decisions
1 reply
0 recast
0 reaction

Chris Carlson pfp
Chris Carlson
@chrislarsc.eth
We need better user interfaces https://farcaster.xyz/chrislarsc.eth/0xb185f440
1 reply
0 recast
0 reaction

androidsixteen pfp
androidsixteen
@androidsixteen.eth
I don't think we'll get them AI is a hail mary at prolonging a "big data" tech paradigm This magic trick will work for a while (and for those who can wield tools deftly), but for genpop, I don't think we'll have any better UIs Where there can be better UIs, this tool will just be integrated into existing products and be called "software" > There’s an old joke that ‘AI’ is whatever doesn’t work yet, because once it works, people say ‘that’s not AI - it’s just software’. https://www.ben-evans.com/benedictevans/2024/5/4/ways-to-think-about-agi
1 reply
0 recast
1 reaction

bence pfp
bence
@bence
will have to read the article later, thanks for sharing on ui ux – i think it can be part of the solution, and in insolation does work that way in context, there is way more money in creating useless over-reliance on the software so if the profit motive is not there, can’t see it happen (hoping to be proven wrong haha)
0 reply
0 recast
1 reaction