Just making it less resource intensive, faster and cheaper to use, will allow way more applications. Hallucinations will be hugely reduced in practice, agents will be less lazy, it will be possible to make it less error prone, because you can have more layers of control and follow-up, and are able to make agents learn from experiences, simply because you're able to provide documentation as it experiences things, improving instructions for them. It will also make it more feasible to create layers for testing out agents and instructions continuously before deploying them. Just in that regard, claiming that it is at its peak doesn't make sense. That's basically ignoring most of what it can be used for.

Yea, I thought this was kind of standard procedure. Since they get lazy, leave stuff unfinished, get facts wrong and hallucinate. You can just get rid of most that by having an AI agent handling another bot, of course. I made one this week that are writing plays with some known manuscript writing methods, you can of connect them to bash shells to make personal assistant - with assignments in e.g. MOTD or that get handle by running commands, and then you can kind of build an AI company with hierarchies, which people already have done of course.

Yes, it is nuts that they use gpt-3.5-turbo and why this paper gets any attention here. Slashdot sucks.

Yea, this seems extremely relevant. Really strange that they weren't clearer.

What's the over all purpose of this study since they use GPT-3.5-Turbo?
At the HumanEval benchmark for coding for example, GPT-4 scores about 40% better than GPT-3.5.
Other prompting can improve the answers further. Using some method called Reflexion, GPT-4 scores about 36% than it does normally on that same measure.
Using Tree of Thought prompting can improve answers by an insane amount.

Looking at the study quickly, it doesn't make sense to draw any bigger over all conclusion about the usability of GPT.
It seems like Slashdot sucks if this is suppose to be worth noticing here.

The whole community here seems generally really behind in some ways.

Have I missed out on something here?

A news service that only works for good looking people.

Excellent link! Should have been upvoted!

Can't draw that conclusion based on nothing

What are you talking about?

1) Wear suspenders, thick glasses, put asthma spray and pens in breast pocket, tuck in shirt, pull pants up high, trip a lot and speak in a weird voice. 2) Occasionally diverge the conversation into unnecessarily detailed subjects you seem to obsess over. 3) Use weird mannerisms. 4) Use a go pro to record the interaction (will look natural if you play your part correctly). 5) Hire a focus group to evaluate the result.

Instead

Didnâ(TM)t read the study, but itâ(TM)s pretty important to know what could be related to pollutants, and what could be related to other factors of urban living. Urbanization has already been known to affect mental health negatively. Hereâ(TM)s one study: https://www.ncbi.nlm.nih.gov/p... But just Google it. Some random citation: âoeThe meta analysis by Reddy and Chandrashekhar(1998) revealed higher prevalence of mental disorders in urban area i.e., 80.6%, whereas it was 48.9% in rural area. Mental disorders primarily composed of depression and neurotic disorders.âoe

Fanny & Alexander, The Elephant Man, Tillsammans, Låt den rätte komma in (Let the right one in, the Swedish original), Mies vailla menneisyyttä (The man with out a past), Brazil, Black Cat/White Cat. I think El Orfanato and Rec (the Spanish horror movie) are good, but more entertaining than good. Lot's of good Tarkovskij ones - I like Stalker and The Sacrifice (Offret). Almodóvar has a lot of good ones. Un Chien Andalou is great, but not underrated I guess.

What kind of nootropics did they take?

Hi, for anyone interested. I've finally come up with a patented method for Noise Amplifying Headphones (NAH). If you want to dull-out uninteresting music from your head phones, this is for you.

Well, it's still a couple years into the future that this will be available for the masses. We only have a 10 kg prototype so far. So unfortunately you'll have to hold out a bit further until I turn this dream into reality.