Meta's AI culture

Setting the record straight on the leaked document

My name has recently hit some news feeds because of an internal document I wrote while in Meta’s GenAI organization. My writing seems to have leaked to a paywalled tech tabloid called ‘The Information’, which painted my narrative close to a raging 'mic drop' before leaving the company.

This could not be further from the truth.

I’m writing this to set the record straight, and give the proper context to this journalistic misrepresentation.

Last Monday, I posted an essay called ‘Fear the Meta culture’ in Meta’s AI organization chat channel. The intent of this essay was to, constructively, help improve the culture within the team. Over the last couple of months, the more I talked to people, the more I felt things were going off the rails. Many people felt disheartened, overworked, and confused. As an organizational psychology and management enthusiast, and after talking to countless of people, I had an idea or two about what was going on. Talking this over with many employees, they urged me to write my ideas down to help improve the internal culture. While hiking to the summit of Mt. Fuji in Japan during my holidays last week, I finally assembled all the ideas from myself and others into a coherent whole, before writing it all out on the plane back to San Francisco.

The most practical place to write an essay

I did not expect that it would blow up. I posted my 9-page essay in the internal AI org chat group, with the idea that perhaps a handful of people would have the patience to read it. Perhaps even a small discussion would occur! Hopefully, I thought, some positive change would come of it. Little did I know a storm would ensue.

Everyone in the Meta AI leadership read it. Several of them reached out to me and said they loved reading it, thought it was well written, and agreed with a lot of the points. They actively shared it with many others internally in the leadership as well.

It also resonated with many employees. The document spread like wildfire within the company. Within a couple of days, it seems everyone had read it and talked about it. Access to the document was limited to the AI organization only, but copies of it were avidly being traded on the lunchroom floor throughout the company. On my last day, people I never met before came up to me to take pictures with me before I left. It had struck a chord.

Most heartwarmingly, more than 100 employees reached out to me in support of my words. The most common feedback I received was that people finally felt heard. Something was not clicking for them, and my essay helped them understand what was going on. It put words to their feelings. My sentiments were shared. Several people that reached out to me felt overworked and blamed themselves. They told me the article helped them realize they were not the only ones suffering. I am happy I was able to reach so many people positively.

Does that sound like a piece of pure smut someone writes out of wrath when leaving the company? Yeah, I didn’t think so.

As you can perhaps tell from this article, I love a bit of cavalier language, have a healthy dose of Dutch directness, and I don’t eschew a nice pun like ‘metastatic’. The article being internal only, I gave myself some creative freedom to be expressive and capture everyone’s attention. Prime quotes usually help people with retention of what they read, and elicit a reaction that leads to discussion. Sadly, some of my poetic, and perhaps cruder, sentences made far too easy quotes to take out of context by the media.

So what did I actually write about?

As you might imagine, I never intended for this to get leaked. It had quite a bit of confidential information in it, so I can only describe it at a high level.

The article had a spicy meatball intro to get a reader’s attention, which is where the quotes are from. However, what I subsequently described were aspects of the MetaAI culture that I believe were unhealthy in the long run for the employees, for the creativity and productivity of our teams, and for the success of the company as a whole. The core aspects being:

• The fear that people feel on a daily basis of reviews and getting fired, commenting on a lack of safeness people felt, which is crucial for morale.

• The necessity of a culture and processes that enable ‘big projects’ to come to fruition.

• The management culture not promoting a sense of camaraderie among employees, leading to a lack of sense of belonging.

• Instability in team assignments, leading to experience not building up and crystallizing over time.

• A wavering vision that was tough for team members to enthusiastically rally behind.

I wrote this with care, citing books, articles, and other sources from behavioral and organizational psychology, organizational research, and history. I provided internal examples of the culture that I was seeing manifested, and at the end provided a list of improvements the company could make to strengthen the culture.

Meta’s leadership agreed with a lot of my points, and many liked the essay. They told me they were aware of many of the issues I addressed, and they were already working actively to change things around. Eager to change things around after the recent top-talent hires. I believe a lot of the issues that arose can be ascribed to the organization running on fumes in a protracted and combative race to the top in AI, and a hardcore software culture that made Meta big and famous, but does not translate very aptly to the world of creating and productionizing LLMs.

So there you have it. No frantic ragequit moment. No arrogant disgruntled engineer leaving while firing a red shell in his rearview mirror. No entitled wizkid bombing his own career by leaking dirty laundry to the press. Just a piece of sensationalist journalism in a time when AI, and especially Meta, is a hot topic, and an employee who tried to improve the world for his colleagues.

Sunrise at Mt. Fuji from the very top

Now that I’m a free spirit once again, I intend to use this substack to share more musings and insights into AI, and my predictions of where the world is heading. I am very passionate about AI and changing the world for the better with new technology. Hopefully I can share some of this passion with the community!

What’s next? I was already planning to leave Meta for some time, and finally pulled the trigger. I am actively job searching, and thinking of new startups to create. Attending ICML this week, hope to see a lot of you there! Let me know if you want to have a chat!