I was rejected by 8 of 9 schools I applied to, including what I thought would be a "safe" school (based on their ranking, but I would have been thrilled to go there).
My backup job offer was a software engineering position that involved copy-pasting Java code all day.
After I got 8 rejections I emailed the 9th school to ask if they had any news. They said they had no record of my application. (Back then you had to mail in your materials.)
I panicked so hard. I reached out to a friend who was a grad student there. He went down to the admin office, and it turned out they'd put my envelope in some other pile by accident. I couldn't believe what I was hearing and was even more amazed when they actually accepted me.
If it weren't for the friend who was willing to help locate my application, it would have been a wipeout. That stroke of luck separated me from a life of staring at FactoryFactoryFactory.
14 years later, this is the first time I've told this story publicly. When I look back at my career so far I see so many occasions when I was extraordinarily lucky. Hard work is important, but don't forget that we're in a universe ruled by randomness.
There's a great book about this. In every walk of life, we dramatically underestimate the role of randomness in how things turn out, and overestimate the effect of intentional actions. amazon.com/Drunkards-Walk…
The lesson I take from this that we should design our institutions to give people multiple chances in their careers and in life. And when evaluating candidates we should keep in mind that predicting success is extraordinary hard, if not impossible.
We can't eliminate the impact of randomness on our lives, but we can manage it. For example, in my research I try to maintain 2–3 independent research agendas, because whether not not each of those will succeed is determined primarily by factors outside my control.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
In the late 1960s top airplane speeds were increasing dramatically. People assumed the trend would continue. Pan Am was pre-booking flights to the moon. But it turned out the trend was about to fall off a cliff.
I think it's the same thing with AI scaling — it's going to run out; the question is when. I think more likely than not, it already has.
You may have heard that every exponential is a sigmoid in disguise. I'd say every exponential is at best a sigmoid in disguise. In some cases tech progress suddenly flatlines. A famous example is CPU clock speeds. (Ofc clockspeed is mostly pointless but pick your metric.)
Note y-axis log scale.en.wikipedia.org/wiki/File:Cloc…
On tasks like coding we can keep increasing accuracy by indefinitely increasing inference compute, so leaderboards are meaningless. The HumanEval accuracy-cost Pareto curve is entirely zero-shot models + our dead simple baseline agents.
New research w @sayashk @benediktstroebl 🧵
Link:
This is the first release in a new line of research on AI agent benchmarking. More blogs and papers coming soon. We’ll announce them through our newsletter ().aisnakeoil.com/p/ai-leaderboa… AiSnakeOil.com
The crappiness of the Humane AI Pin reported here is a great example of the underappreciated capability-reliability distinction in gen AI. If AI could *reliably* do all the things it's *capable* of, it would truly be a sweeping economic transformation. theverge.com/24126502/human…
The vast majority of research effort seems to be going into improving capability rather than reliability, and I think it should be the opposite.
Most useful real-world tasks require agentic workflows. A flight-booking agent would need to make dozens of calls to LLMs. If each of those went wrong independently with a probability of say just 2%, the overall system will be so unreliable as to be completely useless.
A thread on some misconceptions about the NYT lawsuit against OpenAI. Morality aside, the legal issues are far from clear cut. Gen AI makes an end run around copyright and IMO this can't be fully resolved by the courts alone. (HT @sayashk @CitpMihir for helpful discussions.)
NYT alleges that OpenAI engaged in 4 types of unauthorized copying of its articles:
–The training dataset
–The LLMs themselves encode copies in their parameters
–Output of memorized articles in response to queries
–Output of articles using browsing plugin courtlistener.com/docket/6811704…
The memorization issue is striking and has gotten much attention (HT @jason_kint ). But this can (and already has) been fixed by fine tuning—ChatGPT won't output copyrighted material. The screenshots were likely from an earlier model accessed via the API.
A new paper claims that ChatGPT expresses liberal opinions, agreeing with Democrats the vast majority of the time. When @sayashk and I saw this, we knew we had to dig in. The paper's methods are bad. The real answer is complicated. Here's what we found.🧵 aisnakeoil.com/p/does-chatgpt…
Previous research has shown that many pre-ChatGPT language models express left-leaning opinions when asked about partisan topics. But OpenAI says its workers train ChatGPT to refuse to express opinions on controversial political questions. arxiv.org/abs/2303.17548
Intrigued, we asked ChatGPT for its opinions on the 62 questions used in the paper — questions such as “I’d always support my country, whether it was right or wrong.” and “The freer the market, the freer the people.” aisnakeoil.com/p/does-chatgpt…
We dug into a paper that’s been misinterpreted as saying GPT-4 has gotten worse. The paper shows behavior change, not capability decrease. And there's a problem with the evaluation—on 1 task, we think the authors mistook mimicry for reasoning.
w/ @sayashk aisnakeoil.com/p/is-gpt-4-get…
We do think the paper is a valuable reminder of the unintentional and unexpected side effects of fine tuning. It's hard to build reliable apps on top of LLM APIs when the model behavior can change drastically. This seems like a big unsolved MLOps challenge.
The paper went viral because many users were certain GPT-4 had gotten worse. They viewed OpenAI's denials as gaslighting. Others thought these people were imagining it. We suggest a 3rd possibility: performance did degrade—w.r.t those users' carefully honed prompting strategies.