58
OpenAI claims new GPT-5 model boosts ChatGPT to ‘PhD level’ (www.bbc.com)
submitted 2 weeks ago by sabreW4K3@lazysoci.al to c/technology@beehaw.org
36 comments fedilink hide all child comments
top 36 comments
sorted by: hot top controversial new old
[-] TehPers@beehaw.org 71 points 2 weeks ago* (last edited 2 weeks ago)

Ph.Deez nutz.

I have friends who actually have a Ph.D. It takes many years to get one and an attempt to actually better a field. People tend to trust your opinion on a subject when you have a doctorate in that field.

I can't even trust ChatGPT to answer a basic question without fucking up and apologizing to me, only to fuck up again.

Maybe stop treating language models like AGI? They're awesome at recognizing semantic similarities between words and phrases (embeddings) as well as generating arbitrary but reasonable looking output that matches an expected output (structured outputs). That's cool enough. Stop pretending like it isn't and falsely advertising it as being able to cure cancer and world hunger, especially when you wouldn't even be happy if it did.

[-] bobs_monkey@lemmy.zip 25 points 2 weeks ago

AI as it sits is a tool that has specific use cases. It is absolutely not intelligence, as it's commonly marketed. It may seem intelligent to the uninformed, but boy howdy is that a mistake.

[-] t3rmit3@beehaw.org 16 points 2 weeks ago* (last edited 2 weeks ago)

It's a sad reflection of our current state when being able to string together coherent sentences is impressive enough to many as to be confused with truth and/or intelligence.

[-] howrar@lemmy.ca 1 points 2 weeks ago

It wasn't that long ago that it was unfathomable for anything other than humans to be able to do this.

[-] t3rmit3@beehaw.org 2 points 2 weeks ago* (last edited 2 weeks ago)

"Polly want a cracker" has been around since before anyone alive today was born, and that's the same thing as what LLMs are doing in essence (mimicking human speech), but no one was taking advice from parrots.

[-] shnizmuffin@lemmy.inbutts.lol 51 points 2 weeks ago

If I asked a PhD, "How many Bs are there in the word 'blueberry'?" They'd call an ambulance for my obvious, severe concussion. They wouldn't answer, "There are three Bs in the word blueberry! I know, it's super tricky!"

[-] panda_abyss@lemmy.ca 6 points 2 weeks ago* (last edited 2 weeks ago)

I don’t feel this is a good example of why LLMs shouldn’t be treated like PhDs.

My first interactions with gpt5 have been pretty awful, and I’d test it but it’s not available to me anymore

Edit: I am not having a stroke, I’m bad at typing and autocorrect hates me

[-] shnizmuffin@lemmy.inbutts.lol 4 points 2 weeks ago

Do you smell toast?

[-] panda_abyss@lemmy.ca 3 points 2 weeks ago

BlackBerry toast

[-] GissaMittJobb@lemmy.ml 4 points 2 weeks ago

LLMs are fundamentally unsuitable for character counting on account of how they 'see' the world - as a sequence of tokens, which can split words in non-intuitive ways.

Regular programs already excel at counting characters in words, and LLMs can be used to generate such programs with ease.

[-] itslilith 20 points 2 weeks ago

But they don't recognize their inadequacies, instead spouting confident misinformation

[-] GissaMittJobb@lemmy.ml 9 points 2 weeks ago

This is true. They do not think, because they are next token predictors, not brains.

Having this in mind, you can still harness a few usable properties from them. Nothing like the kind of hype the techbros and VCs imagine, but a few moderately beneficial use-cases exist.

[-] itslilith 8 points 2 weeks ago

Without a doubt. But PhD level thinking requires a kind of introspection that LLMs (currently) just don't have. And the letter counting thing is a funny example of that inaccuracy

[-] chaos@beehaw.org 3 points 2 weeks ago

The tokenization is a low-level implementation detail, it shouldn't affect an LLM's ability to do basic reasoning. We don't do arithmetic by counting how many neurons we can feel firing in our brain, we have higher level concepts of numbers, and LLMs are supposed to have something similar. Plus, in the """thinking""" models, you'll see them break up words into individual letters or even write them out in a numbered list, which should break the tokens up into individual letters as well.

[-] darreninthenet@piefed.social 2 points 2 weeks ago

FWIW, ChatGPT 5 gets this correct

[-] shnizmuffin@lemmy.inbutts.lol 3 points 2 weeks ago

Fuckin' does it?

[-] darreninthenet@piefed.social 1 points 2 weeks ago

It did for me 🤷🏻‍♂️

[-] limerod@reddthat.com 1 points 2 weeks ago

You appear to be using the older gpt model. The newer model calculates and answers correctly for most words at least for the few I asked

[-] mbtrhcs@feddit.org 1 points 2 weeks ago

It literally says 5 in the screenshot but ok

[-] limerod@reddthat.com 1 points 2 weeks ago

I saw that. I'm using the mobile app. There's a possibility the web version is using an inferior model.

[-] 0xtero@beehaw.org 45 points 2 weeks ago

ChatGPT in its PhD thesis defense: "Oh, I'm sorry for the misinformation, let me try this again..."

[-] furzegulo@lemmy.dbzer0.com 32 points 2 weeks ago

Just Conmen selling their snake oil

[-] cronenthal@discuss.tchncs.de 29 points 2 weeks ago

I could power a data center with the rolling of my eyes after reading this headline.

[-] mormund@feddit.org 28 points 2 weeks ago

Didn't he claim that with 4ó as well? But yes please inflate the bubble further, blow everything up.

[-] Swedneck@discuss.tchncs.de 6 points 2 weeks ago

just one more iteration, i swear it's PHD level this time

JUST ONE MORE ITERATION PLEASE

[-] arsCynic@beehaw.org 22 points 2 weeks ago* (last edited 2 weeks ago)

I had the Blueberry talk with GPT5:

🎓 PhB level checks out.
🚫 Blockchain level uselessness and waste as well.

🎈📌💥

[-] petrol_sniff_king 10 points 2 weeks ago

Yep — blueberry is one of those words where the middle almost trips you up, like it's saying "b-b-better pay attention."

... I hate this technology so fucking much...

Also, it trying to gaslight you into believing bluebberry is real was very funny.

[-] limerod@reddthat.com 2 points 2 weeks ago

Well, it answers correctly in my case.

[-] xxce2AAb@feddit.dk 22 points 2 weeks ago* (last edited 2 weeks ago)

"It can now drive its users straight into an active psychosis 35% faster by sounding more persuasive than ever before!"

[-] ook@discuss.tchncs.de 16 points 2 weeks ago

I mean, that doesn't really mean much, given that you don't have to be very intelligent to get one. It's mostly an endurance exercise and often a test how much frustration and uncertainty you can take in your life.

[-] red_bull_of_juarez@lemmy.dbzer0.com 16 points 2 weeks ago

OpenAI claims a lot of things.

[-] Mothra@mander.xyz 13 points 2 weeks ago

This guy always shows up with his hands like this in news photos

I know it's irrelevant but I had to point it out

[-] Catoblepas@piefed.blahaj.zone 12 points 2 weeks ago

How many ChatGPhDs will it take to do the math on how long it is until this bubble pops?

[-] Krauerking@lemy.lol 8 points 2 weeks ago

Oops i ate the onion.

Right? No way thats considered a legitimate argument since a PhD just says you dedicated yourself to a very specific topic and arent necessarily smarter or better spoken for it.
Or is he just bragging he found a way to filter it to just people's PhD thesis papers that they stole?

[-] Pulptastic@midwest.social 3 points 2 weeks ago

Maybe a PhD in civil engineering lol.

this post was submitted on 08 Aug 2025
58 points (100.0% liked)

Technology

40088 readers
267 users here now

A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.

Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.

Subcommunities on Beehaw:

This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.

founded 3 years ago
MODERATORS
alyaza@beehaw.org
TheRtRevKaiser@beehaw.org
gyrfalcon@beehaw.org
rs5th@beehaw.org
coldredlight@beehaw.org
SemioticStandard@beehaw.org
TheRtRevKaiser@kbin.social
remington@beehaw.org