Alexa vs OpenAI
Are OpenAI devs really that good they can outperform 15k devs or what the duck is happening in the Alexa org? A lot of these devs in Alexa org went to top schools or the scientists likely have PhDs… Who is making the decisions? We have that AlexaTM model and the Codewhisperer one which aren’t as good as GPT but I mean we have someone working on models so why the duck can 15k people not make them as good as GPT? Oh and the kicker is we have the most cloud compute in the world and even our own DL accelerators… we don’t need to sell out to Microsoft…
More engineers does not mean better product
I work in Alexa ai with some of those scientists. They aren’t as good as they look and aren’t really doing true research. They’re just taking already done research and applying it to Alexa.
We hire the fakers with phds, not the actual experts. Which good scientist would stay working on Alexa?
Nine women won’t deliver a baby in a month.
They’re good, but honestly they’re probably a little bit better than Alexa research people (as far as raw talent. Everything else OpenAI wins) Difference is they’re lean, can ship quickly, and engineers are incentivized with huge pay packages and equity (seriously, go look at levels.fyi for open ai it will blow your fucking mind) Amazon enforces the exact opposite. Bloat from empire building, pay packages with minimal equity, etc On Twitter some dude was saying OpenAI consistently has engineers staying until 1 or 2am. That’s how you win
Lol what? Openai is way better than amazon
Yeah you’re right. I was just saying among research scientists maybe IQ wise the average at both is probably not drastically different (SDEs it is drastically different) but what I’m trying to say is the IQ doesn’t matter as much as the fact that everyone there is grinding and working hard, where Amazon doesn’t incentivize that
ITT: Designers blaming PMs. PMs blaming engineers. Engineers blaming scientists and ultimately everyone blaming leadership. If it isn't already clear with the layoffs, a single large monolith will eventually struggle at scale. Amazon will be the next IBM, Meta the next Sun and so on. Execution speed bottles down to a team of people following a single vision, regardless of size. OpenAI will eventually go through this phase too.
Even in a large company, progress is possible with competent leadership and clear vision. In Alexa we have a bunch of overpaid snakes focused on filling their pockets, building empires and avoiding PIP
Like I said everyone blames leadership at the end of the day. One poor decision maker for a key product and the house of cards come tumbling down
Grand ma vs Hot chick
Corporate culture matters. Pip machine style culture may work for real factories but it doesn't work in innovation.
word of wisdom
How much cash is Amazon sitting on? We spent a loooot on just training the model.
Likely because of those useless and annoying PMs getting in the way ..