@moocat999 @ksorbs @grok Who are you and why didn't you watch the full clip to understand that his point actually was the complete opposite of what you think?
English
Kasper Saugmann
8.1K posts
Kasper Saugmann
@kaspers
Tech journalist and entrepreneur. Words in @borsendk @frihedsbrevet @weekendavisen @videnskabdk, and more. Former co-founder @denuafhaengige (acq)
Copenhagen, Denmark เข้าร่วม Mayıs 2008
1.1K กำลังติดตาม1.5K ผู้ติดตาม
I still can't this is real
Q: “The Promised Land” is entirely Nordic and it lacks diversity.
Arcel: “Hmm.. well first of all, the film takes place in Denmark in the 1750s”
English
@scmallaby @sama @demishassabis Source for saying that Altman's primary (sole?) motivation is power?
English
What motivates the creators of AI? @sama wants power. Zuck wants commercial success. @demishassabis is motivated by a scientific curiosity so intense as to be spiritual. Read my biography of Demis, out March 31. THE INFINITY MACHINE.
a.co/d/0aHD3OfQ
Demis Hassabis@demishassabis
It’s 2am again, my favourite time, and as always reality is still staring me in the face 🧐
English
@Jens57348198 @ammitzbollbille @PeterMAstrup Hvir meget vil du vurdere, at det i så fald vil løbe op i pr. dansker? 30 øre?
Dansk
@ammitzbollbille @PeterMAstrup Måske fordi danske skatteydere også kommer til at betale for hendes promovering til sådanne ansøgninger - lige som de gjorde i 2023-2024.
Derfor er det naturligvis relevant.
Dansk
Hvis jeg stadig var journalist og skulle med til Socialdemokraternes pressemøde kl. 12:15, så ville jeg spørge om:
- Hvis du bliver statsminister igen kan du så garantere, at du ikke forlader posten til fordel for et topjob i udlandet?
- Meget tyder på, at der var meget begrænset og måske slet ingen droner. Hvorfor sagde du, at vi var “under angreb” fra en “kapabel aktør”?
- Du har i adskillige år sagt, at danskerne skal arbejde mere, fordi russerne ikke går hjem fra samlebåndet og det hele ikke “hænger sammen” ellers, men nu tre dage inden valget, så skal danskerne pludselig have ret til deltid? Wat?
Dansk
I am now at 5 GPU providers being completely sold out for a single node of 8xH100s.
I don’t think people understand the gravity of what is about to come.
Suhail@Suhail
The run on inference capacity is coming. You have been warned.
English
@HoffPeterA @pvergadia Half of the participants in the study didn't use AI
English
JUST DROPPED: Anthropic's research proves AI coding tools are secretly making developers worse.
"AI use impairs conceptual understanding, code reading, and debugging without delivering significant efficiency gains." -- That's the paper's actual conclusion.
17% score drop learning new libraries with AI.
Sub-40% scores when AI wrote everything.
0 measurable speed improvement.
→ Prompting replaces thinking, not just typing
→ Comprehension gaps compound — you ship code you can't debug
→ The productivity illusion hides until something breaks in prod
Here's why this changes everything:
Speed metrics look fine on a dashboard.
Understanding gaps don't show up until a critical failur and when they do the whole team is lost.
Forcing AI adoption for "10x output" is a slow-burning technical debt nobody is measuring.
Full paper: arxiv.org/abs/2601.20245
English
@drapersgulld I love Claude Code and it was actually my first choice for the task, but after it failed miserably for ten minutes in fetching simple data, I gave Lovable a shot and it worked right away.
English
The trick is what happens after the 10 minutes.
Kasper Saugmann@kaspers
@AdamBartas I just used Lovable and Claude Code for the same task. CC scrambled for ten minutes, Lovable just worked
English
@pvergadia Why not say both points out loud? The pitfalls and how to achieve enhanced performance?
English
I work with enterprises and unfortunately <10% users actually use these tools in efficient ways where they generate and comprehend. 90% use them like another ChatGPT or Claude with bad prompts! After more than 6 months of training and comprehensive change management we get maybe 30-40% to a good usage pattern.
This is exactly the point that needs to be said out loud because the reality on ground in real enterprises is very different from what we read on social media!
These are real humans trying to work in an entirely different way, building new habit take effort and time and lots of leadership support.
Kasper Saugmann@kaspers
@pvergadia Nah, dropped six weeks ago, and the your conclusion is more nuanced than that. It depends on how you use it. If you use it correctly (Generation-Then-Comprehension) then it will yield even better results than without. Didn't you read the paper?
English
@AdamBartas I just used Lovable and Claude Code for the same task. CC scrambled for ten minutes, Lovable just worked
English
First sign that Lovable is dead
Pivoting to general assistant is the most "investor-pleasing" move you could do
Their app building business is obv going nowhere and investor money is drying up
Why should anyone use Lovable instead of the already established ecosystems
Anton Osika – eu/acc@antonosika
Introducing Lovable for more general tasks. Lovable has always been for building apps. Today it also becomes your data scientist, your business analyst, your deck builder, and your marketing assistant. This is a big step toward what Lovable is becoming: a general-purpose co-founder that can do anything. See examples below.
English
@kaspers @pvergadia Is that why companies are laying off tens of thousands of engineers and mandating the use of generative coding tools, causing major outages at hyperscalers? Because people are taking the time to review these outputs?
English
@nyc_data_geek @pvergadia Yes, some clearly are, that's why it shows up in the study ... and more should.
English
@kaspers @pvergadia No one is doing that though. Comprehension demands taking time to understand. The whole point is 10000000x output. No one is taking time to comprehend
English
@GaryMarcus It depends on how you use it, as the paper says.
The "Generation-Then-Comprehension" outperforms all other uses, including no usage.
Didn't you read the paper?
English
“AI use impairs conceptual understanding, code reading, and debugging without delivering significant efficiency gains” — and that’s a quote from Anthropic’s own research!
Priyanka Vergadia@pvergadia
JUST DROPPED: Anthropic's research proves AI coding tools are secretly making developers worse. "AI use impairs conceptual understanding, code reading, and debugging without delivering significant efficiency gains." -- That's the paper's actual conclusion. 17% score drop learning new libraries with AI. Sub-40% scores when AI wrote everything. 0 measurable speed improvement. → Prompting replaces thinking, not just typing → Comprehension gaps compound — you ship code you can't debug → The productivity illusion hides until something breaks in prod Here's why this changes everything: Speed metrics look fine on a dashboard. Understanding gaps don't show up until a critical failur and when they do the whole team is lost. Forcing AI adoption for "10x output" is a slow-burning technical debt nobody is measuring. Full paper: arxiv.org/abs/2601.20245
English
@richard_uw @antiscarcity @sama I hope you're not counting those ZIRP levels as any kind of normality
English
English
I have so much gratitude to people who wrote extremely complex software character-by-character. It already feels difficult to remember how much effort it really took.
Thank you for getting us to this point.
English
@sama You're welcome. Nice to know that our reward is our jobs being taken away.
English
@pelledragsted Hvis du mener at være bedre til at drive bank end Nordea, skylder du så ikke næsten samfundet at tage det sats, det er at starte noget nyt og vise dem, hvor skabet skal stå?
Dansk
Det er jo barokt.
Gigantbanken Nordea havde sidste år et overskud på over 30 milliarder. Nu nedlægger de 1500 job.
Det er bl.a. den virksomhed, Venstre vil forgylde med deres økonomiske plan.
En skattelettelse på 252 millioner - en kvart milliard - mener Venstre, Nordea fortjener.
Jeg vil hellere støtte arbejderen, pensionisten og børnefamilien end storbankerne. #dkpol
dr.dk/nyheder/senest…
Dansk
Misleading summary. Should be deleted.
Altman doesn’t say a (known) new architecture is coming; he says he anticipates one will come someday.
PS: I also think we need something radical and new. In fact that’s what I’ve been saying for the last decade.
Excess focus on exploiting LLMs has likely delayed discovery.
Rohan Paul@rohanpaul_ai
Sam Altman just said in his new interview, that a new AI architecture is coming that will be a massive upgrade, just like Transformers were over Long Short-Term Memory. And also now the current class of frontier models are powerful enough to have the brainpower needed to help us research these ideas. His advice is to use the current AI to help you find that next giant step forward. --- From 'TreeHacks' YT Channel (link in comment)
English
Martin Smith@martinalexsmith
@EganPeltan Canine checkpoint inhibitor + custom mRNA vaccine led to the drastic reduction of tumours. Not sure I can say more than neoantigens were validated by sequencing, for now. We are also performing additional cellular and molecular tests to check specificity of the immune response.
QME
Ok this is ridiculous. Everything here could have been done without ChatGPT
1. The dog is on conventional immunotherapy with the mRNA vax
2. It appears the mRNA vaccine started WITH ICI, so we can’t know if the vax had ANY additional effect
3. The team can’t say what ~~AI~~ identified the neoantigens (no, not AF3). It sounds like they used existing sequence homology workflows.
4. No evidence of antigen-specific effect from mRNA vax - the authors need to prove this before anyone can believe the neoantigen selection had any effect
5. The in-kind contributions here are ~$20-50k. Custom cancer vax isn’t cheap
Custom mRNA cancer vaccines have been in development for years! None have been clear, resounding successes (yet). Once we have Phase 3 PFS/OS, not N=1 anecdotes, we can start having arguments about people being denied access to effective treatments by unnecessary regulation
Right now, the only thing people are denied are the opportunity to fork over $50k for hope and dreams. Should we really make it easier for people exploiting desperation to sell unproven remedies?
If you don’t want to think before you tweet, ask the fucking AGI if any of these claims are remotely plausible and what evidence you’d need to believe them
vittorio@IterIntellectus
this is actually insane > be tech guy in australia > adopt cancer riddled rescue dog, months to live > not_going_to_give_you_up.mp4 > pay $3,000 to sequence her tumor DNA > feed it to ChatGPT and AlphaFold > zero background in biology > identify mutated proteins, match them to drug targets > design a custom mRNA cancer vaccine from scratch > genomics professor is “gobsmacked” that some puppy lover did this on his own > need ethics approval to administer it > red tape takes longer than designing the vaccine > 3 months, finally approved > drive 10 hours to get rosie her first injection > tumor halves > coat gets glossy again > dog is alive and happy > professor: “if we can do this for a dog, why aren’t we rolling this out to humans?” one man with a chatbot, and $3,000 just outperformed the entire pharmaceutical discovery pipeline. we are going to cure so many diseases. I dont think people realize how good things are going to get
English