Key talent at OpenAI is leaving the company in protest against a corporate culture that prioritizes commercial performance over safety and unfair exit conditions.
Imagine this: You have just joined your dream company. It’s a place that creates incredible technology destined to change the world. However, over time, the company becomes more obsessed with churning out profitable products quickly than with the value of “safe AI for humanity” you initially believed in. You find yourself torn. “Is it right for me to keep working here?”
Recently, at OpenAI, the “icon” of the AI industry, this very dilemma has led to key talent leaving the company in droves. What on earth is happening inside the world’s top AI company?
Why does this matter?
AI has become inseparable from our daily lives. If the culture of the company building this technology falters, the impact will inevitably fall upon the users. Rushing product releases while sidelining safety can lead to unexpected technical errors or ethical issues. As OpenAI is the company setting industry standards, this “exodus” is read as a dangerous sign that the balance between the speed of technological advancement and safety is collapsing.
Easy to understand: An elite team that forgot the ‘fundamentals’
Shall we compare AI technology development to ‘studying’ in school? Let’s say there is ‘safety and fundamental research’ to gradually build up basic principles to solve math problems, and ‘product release’ to quickly write down the answer and get a score.
Until recently, OpenAI was famous as a team that thoroughly built up fundamental principles. However, the atmosphere has changed completely. A former safety team lead who left OpenAI criticized the company’s culture as having deteriorated into one that “prioritizes product releases over safety” Reference 1, Reference 12.
It is similar to a marathon runner taking risky shortcuts to shorten their record. If they run fast without safety devices, their record might improve, but they could trip and fall badly at any time. In fact, the team responsible for AI safety was only allocated 20% of the total computing resources, and was eventually dissolved a few weeks later Reference 12.
Simply put, it’s like pouring the entire budget into a team that increases engine output to make a car go fast immediately, rather than into a team that researches brakes and seatbelts. Naturally, the technology will develop rapidly, but the risks will increase proportionally.
Current situation: The trap of ‘gag’ agreements
There is also a structural problem that makes it difficult for employees to leave the company. OpenAI requires departing employees to sign an exit agreement that includes a ‘non-disparagement’ clause. If they refuse, they risk forfeiting millions of dollars in their company equity (stock) Reference 3. It is a situation where they must essentially choose between abandoning the rewards of years of hard work or risking a ‘whistleblowing’ exit that criticizes the company. It is tantamount to economic shackles placed on the employees.
Individual cases are also serious. Researcher Zoë Hitzig resigned in protest after OpenAI announced plans to introduce advertisements into ChatGPT Reference 2. Already in November 2023, CEO Sam Altman was dismissed due to communication issues with the board, and co-founder Greg Brockman resigned immediately after, showing a repeated crisis of internal trust Reference 9, Reference 13.
What will happen in the future?
Experts point out that this brain drain will change the landscape of technological competition. However, if OpenAI remains obsessed only with product speed, other companies might take the lead by securing both safety and technological superiority simultaneously.
Currently, many who have left are moving to other competitors or research institutes that value safety more highly Reference 4. Ultimately, OpenAI’s crisis is not just a problem for one company, but is likely to be growing pains as global artificial intelligence technology transitions from ‘quantitative expansion’ to ‘ethical substance.’
By analogy, it is a situation where the ‘AI era,’ a child that has just started to walk, is in danger of falling because it is trying to run too fast. Is it not time for us to check the ‘direction’ rather than the speed so that the child can grow up healthy? Readers, which do you think is more important, ‘fast technology’ or ‘safe technology’?
AI’s Perspective
MindTickleBytes’ AI reporter perspective: It is natural for companies to pursue growth. However, for a company handling a tool with as immense influence as AI, safety is not an option but a reason for existence. To overcome the current crisis, OpenAI needs to break away from its current product-centric thinking and fundamentally reflect on what value its technology brings to humanity.
References
- IQuitOpenAIBecauseItsCultureIsBroken
- The Glut of ‘WhyIQuit’ Letters Is Out of Control - Business Insider
-
[Why theOpenAIsuperalignment team in charge of AI safety… Vox](https://www.vox.com/future-perfect/2024/5/17/24158403/openai-resignations-ai-safety-ilya-sutskever-jan-leike-artificial-intelligence) - HeQuitOpenAI— and Risked Millions to Warn Us - YouTube
-
[Sacking, revolt, return: how crisis atOpenAIover Sam… The Guardian](https://www.theguardian.com/technology/2023/nov/25/how-crisis-openai-sam-altman-unfolded) - Why is everyone leavingOpenAI?
- OpenAIPresident Greg Brockman resigns after Sam Altman ouster
- Salary reduction
- Forfeiture of equity (stock)
- Non-payment of severance pay
- Dissolution of the safety team
- Announcement of plans to introduce ads to ChatGPT
- Dismissal of the CEO
- Poor management performance
- Unethical research
- Lack of candor in communications