AughtDev
@aughtdev
Followers
46
Following
82
Media
119
Statuses
841
I love feeling the population of my TL drop in real time over the weekend. Nice to know that some of you have a life
0
0
0
10 years ago these are the people who told people not to bother studying radiology because a few models could identity cancer on CAT scans. Listen to them at your own risk.
Founder of Google's Generative AI Team Says Don't Even Bother Getting a Law or Medical Degree, Because AI's Going to Destroy Both Those Careers Before You Can Even Graduate. "Either get into something niche like AI for biology. or just don't get into anything at all.". The next
0
0
0
Logos are going to become like musical riffs, where the generic ones are too similar to enforce any sort of copyright.
0
0
0
This has been a predictable outcome for anybody who's tried to use gen AI for complex work. The hype around businesses adopting AI exponentially was just FOMO. Even Satya Nadella said last year that we can't call anything AGI until there's a noticeable global GDP increase.
🚨 Is the "AI first" approach worth it? Should companies expand generative AI deployment at any cost? Here's what the MIT report that is giving Silicon Valley investors extreme ANXIETY said:. "Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a
0
0
0
Much like LLMs, for emotionally salient topics people pretrain themselves into a narrow distribution of opinions and responses that they parrot automatically. I think this is just a case of motivated reasoning, justifying what you already feel without thinking too hard.
So many people are LLMs, it's incredible. You say Macron is a spineless clown, they say you support Putin. You say that altman is a lying jerk, they say you hate AI. You say you cannot discriminate against people by race when hiring or admitting to a uni, they say you are.
0
0
0
This feels like consolidation. Locking in for the long run, not giving up.
Breaking 🫧 News: Meta freezes AI hiring and becomes the first major company succumb to the Great AI Bubble Deflation of 2025.
0
0
0
Claude : "Problem solved, the buggy data has been removed". "Buggy data??"
Claude code just made this dev cry after deleting all PDFs, chats, and user data from the DB 🥲 LMAO. It's all good until they mess it up.
0
0
1
This is definitely notable. I suppose that the LLM training process is enough to bridge short gaps in human knowledge therefore within the vastness of LLM latent space, there's rare islands of useful novelty. Reasoners provide a probabilistic way to search through this latent.
Claim: gpt-5-pro can prove new interesting mathematics. Proof: I took a convex optimization paper with a clean open problem in it and asked gpt-5-pro to work on it. It proved a better bound than what is in the paper, and I checked the proof it's correct. Details below.
0
0
0
UBI enthusiasts in shambles. There's no proof that UBI will make people particularly satisfied even in a post AGI world. We'll use it to support ourselves but beyond that I've always believed we'll just acquire a new set of pathologies and social grievances.
It's a bit painful for me to share this, but I respect @KelseyTuoc (and @evavivalt) a lot, and I think they're basically right: the most rigorous recent studies on basic income and cash transfers (in rich countries) have yielded pretty disappointing results.
0
0
0
In the event of confusion. To oneshot -> perform a task successfully in one attempt. To be oneshotted -> to be significantly affected on first encounter with an action/event/thing
0
0
1
This is just the natural evolution of new industries. You first get the pioneers (high agency risk takers), then the followers (medium agency, risk tolerant) then the strivers treading a well defined path (low agency grinders). There's not many pioneers left in tech startups.
the problem with startup culture is that there are IQ130 ivy grads aspiring to build calorie trackers and ai to-do lists. as tech culture is becoming more famous, we are seeing it being diluted with non-ambitious, low-risk people building chill startups. the whole intellectual
0
0
0
If LLMs are any indication there might be several other steps on the road to AGI where we end up with systems that can mostly accomplish a task that we currently associate with intelligence but are not quite good enough and eventually plateau.
1
0
1
Watching LLMs and other ML based systems fail on such trivial fail cases makes you appreciate how robust animal intelligence is. Even a rat's brain would not get confused by shifting a few pixels in an image.
1
0
3
Why would you sell an account with your name on it in a industry where reputation is so valuable? now that account is going round being shady with your name. .
0
0
0
> routing engine that intelligently chooses the right model for the user's needs.> Look inside . > Still needs a phD in prompt engineering to use
GPT-5 represents the first model where we finally get to see the limitations of human intelligence when it comes to the utility of this technology. GPT-5 is an amazing model, if you hear differently, it’s a skill issue.
0
0
2
When you get to 100% market share, you can increase the price to 1.25$ a dollar and make crazy profits. It's genius.
i am starting a company where we sell one dollar bills for 90 cents. we will raise $1B in seed capital. the projected first year’s revenue will be $900m. at a 10x sales multiple, the company will be worth $10B. i’ll sell my stock in secondary sales and get rich. i am very smart.
1
0
1
It must be nice to have close friends.
turned on data after a 5 hour flight only to learn that my house had flooded, my friends had formed a 14 person coordination chat, a shop vac was acquired, all wet items were removed and dried or thrown out, our house has been disinfected. and everything is all good now
0
0
1
