Despite the great joys that Openai and his CEO, Sam Altman, are giving the artificial intelligence sector, there are already several voices that have talked about how the company is from within. The truth is that all these already extrabajadores tell somewhat peliagudas situations.
In a nutshell, inside everything is much less orderly than it seems. This is what Calvin French-Owen tells in a very personal blog, one of his star exingenier, who after a year of authentic madness and sleepless nights, has decided to talk clearly about the «chaos and speed» that reign in the Chatgpt company.
French-Owen, who previously founded Segment and was acquired by Twilio for 3.2 billion dollars, joined OpenAi just when boom of the AI was exploiting.
In its recent departure it told without filters how the company went from 1,000 to 3,000 employees in less than a year, with all that that implies. He talks about improvised processes, products launched to counterreloj and an impossible culture to follow.
Specifically, He explains that the internal rhythm is more like a startup freshly launched to the market than to a multinational with years of experience. «Everything breaks when you grow so fast: communication, reports, how product is launched, how people are managed,» he says.
The result are teams that doubled tasks without realizing and some projects developed «half a dozen versions» of the same tools, just because they could.
But be careful, because the improvisation also reaches the source code and with Google engineers working side by side with university, the central «monolith» of Python ends up being «a landfill» in the words of French-Owen himself. Bugs, systems that were hung every two by three and falls that were (and are) part of the routine.
This is how you live within the madness that is to work in OpenAI
In addition to all this madness to get products at full speed, the other great pillar is media paranoia. French-Owen explains that it is an obsessed company with avoiding leaks and that watches absolutely everything that is said in X. «A friend told me: this company works based on Twitter vibes,» he says.
The company, according to the exingeniero, knows that Chatgpt is one of the most evaluated, controlled and investigated products of the world. Each bug, each error or each controversy becomes a holder that monopolizes media around the world and any user can, seconds, make a failure viral. That has done, he explains, that the culture that is lived internally is very closed and secret.
And not only governments, media and users are attentive. Also competition from Google to Anthropic through startups as a cursor, nor do he lose sight of OpenAI. The pressure is maximum because «the rivals are growing just as fast and have excellent engineers,» he explains.
But they also give misunderstandings and one of the great myths, says French-Owen, is to think that Openai goes from security due to pure ambition. «Internally, there is much more focus on practical risks (hate, abuse, political manipulation, cybersecurity …) than about the great science fiction apocalypse that they like so much.»
«We are less than three years after the catastrophe»
This, Interestingly, it is something that clashes directly with the words of another former employee of this company. William Saunders, who worked for three years in the OpenAI super -alignment team, being one of those responsible for his popular chatgptand that he resigned from his post a year ago, he says that he did it because an irresponsible management of the risks related to AI by the company was being carried out.
The engineer compared his former work with whom they built the Titanic in his day. As with the famous transatlantic, the company has an excess of confidence in its security measures and a lack of preparation for possible catastrophes.
Going more thoroughly with the subject, He believes that within three years we could already suffer the consequences of what, in his way of seeing things, is an irresponsible way of acting. Of course, nothing to do with this vision with which other employees have.
What all of them coincide is that all this speed and pressure has effects and the talent that comes from companies such as Google, goal or even new startups, finds an environment where learning is continuous, but exhaustion also and that certainly takes its toll.
Know How we work in NoticiasVE.
Tags: Artificial Intelligence, Work, Software