In this era of AI competition in full swing, $Meta Platforms, Inc.(META)$ on the one hand, high-profile money (recently rumors of a succession of nine-figure high-tech poached $Apple(AAPL)$ AI responsible and OPENAI scientists), the formation of a super team, but on the other hand, it was revealed that the internal culture of the "spread like cancer!"We have to let people start to think seriously: a lack of clear direction, a lack of trust and collaboration of the organization, in the end, how far can go?
'Culture Collapse' Is Meta AI's Achilles' Heel
This is not alarmist talk.An open email from former Meta AI researcher Tijmen Blankevoort when he left the company was a thunderbolt.He pointed out in no uncertain terms that Meta AI was being eaten away by a "culture of fear" - everyone was worried about underperforming and being laid off at any time, and in order to stay alive, they could only compete for projects, brush up on their presence, and watch out for each other.In this environment, innovation is no longer a priority, but survival.
One researcher openly compared Meta to a "metastatic cancer", which is not dramatic rhetoric, but points out the real pain that this huge organization is experiencing: distrust, non-cooperation, and lack of direction.
Smashing money can buy people, but it can't buy a sense of belonging
Meta's investment in AI talent over the years can be called a "big gamble" - millions of dollars in annual salary is not a legend, but the norm.The problem is that the top talent bought at a high salary has not become part of the organization.People are invited in, but the team is still fragmented, do not collaborate with each other, even Meta's own Ray-Ban glasses program has appeared AI team and Reality Labs completely disconnected from the situation.
This is reminiscent of OpenAI's Altman's quote, "Missionaries will beat mercenaries." (People with a sense of mission will eventually beat people who just get paid to work.)Meta's problem is exactly that - people come in not for the vision, but for the big paycheck, and the company doesn't have a clear enough vision to rally them.
Is terminal elimination a good rule of thumb for operations or not?
Organizational Chaos, Resource Fighting, and Talent Loss Become the Norm
What's even more ironic is that within Meta, which claims to have the strongest AI team in the world, different groups are fighting over computational resources; the Llama project is one of Meta's most promising AI achievements, but a number of key researchers involved in the project have left one after another, one of the reasons being the serious infighting and chaotic allocation of resources.You can't expect a group of people to make breakthrough technology while worrying about not having enough computing power and preventing their teammates from "taking credit".
In other words, Meta does not lack of funds, not lack of people, the real lack of a can unite people, so that they are willing to work together.
Want to turn around, Meta should first cure "internal injuries".
It is gratifying that after the exposure of this internal e-mail, Meta executives did not just cover up, but quickly responded.They said they would reflect on the situation and try to reshape the AI culture by bringing in new executives and creating a "Superintelligence Lab".This is good, of course, but the real challenge has just begun.
Culture repair is not a slogan, nor can it be solved by promoting a few managers; Meta has to make a real move in the following areas:
Slow down the "KPI-driven + layoff-inspired" oppressive management mechanism, and let the team return to long-termism;
Reorganize the overall AI vision and collaboration mechanism to break down the information barriers and involutional structure between departments;
Establishing a fair and transparent resource allocation and incentive mechanism, so that technicians can do research at ease instead of infighting.
The AI competition is not just about GPUs, but also about organizational vitality.
In today's AI track, it's not just about who has stronger models and more chips, it's also about who can build an environment where smart people are willing to stay and really collaborate.From this point of view, if Meta can't face up to the internal problems, even more Llama, even stronger models, may be just a flash in the pan.
It is important to realize that it is never the algorithms that really drive AI progress, but the creativity that can grow in trust and freedom.
Meta has to heal itself, now, if it still wants to win the future in the AI war game.
Comments