Do you really need a PhD to do great research?

The OpenAI team has come up with a compelling answer -

'No need!

Four of OpenAI's most famous members, Christopher Olah, Alec Radford, Jason Wei, and Greg Brockman, do not have doctorates.

Christopher Olah, the co-founder of Anthropic, doesn't even have a bachelor's degree, and Greg Brockman dropped out of Harvard and MIT twice to start his own business.

And their impressive achievements, but also can not help but trigger people's thinking -

"Do we need an advanced degree or not?"

No good degrees, but 'good students'

Christopher Olah

At the age of 18, Christopher Olah entered the University of Toronto as an undergraduate. But he soon dropped out of school to study computers, work on open source projects, and intern at companies.

Although he did not have a full college education, and only a high school degree, he learned the latest AI technology through self-study.

Interestingly, the background he wrote for himself on linkedin is "Real Life Experience University."

Christopher Olah worked at Google Brain, was the lead researcher on interpretability in OpenAI, and is currently the co-founder of Anthropic.

Because he didn't have time for school, he became a full-time researcher at Google Brain at the age of 23, and began to mentor and even manage researchers and doctoral interns who were often much older than he was.

Many people email him to ask if college is necessary for personal development.

In response, he published a blog in 2020 called "Do I Need to go to College?"

Blog: https://colah.github.io/posts/2020-05-University/

Instead of asking the question "Should I have a PhD?" he thinks the question should be "Where can I find the best environment to grow as a researcher?"

He also doesn't shy away from the fact that his past experiences tend to be more unconventional --

"I don't really encourage people I don't know very well to take non-mainstream paths, because those paths are more risky and so much depends on the individual."

In the same way, he affirms the risk of not having a degree --

"Not having a degree can have some long-term negative consequences. A college degree is a quick way to show others that you have some basic skills."

Christopher subtly points out that one of the key benefits of college is social interaction, especially when it comes to finding the right mate.

"For many people, university is a time of social development. They learn social skills, make long-term friends and form romantic relationships, which seems to be the biggest benefit of going to university."

"Going to college helps you find your life partner, and missing out on that may be the biggest price you pay for not going to college, but what really matters is getting yourself out there and meeting people who might be in your league."

Alec Radford

Alec Radford works on promising image and text machine learning technologies and moves them from the research stage to developers and wider industry applications.

Alec Radford went to work after receiving his undergraduate degree from Franklin W. Olin College of Engineering.

At the beginning of his career, Radford spent eight years at OpenAI, where he was involved in a number of high-profile AI projects, from GPT-1 to GPT-4, as well as CLIP, DALL·E, etc. In particular, his papers have been cited more than 190,000 times, becoming the industry benchmark.

In particular, his 2020 paper "Language models are few-shot learners" has received 38,456 citations so far.

And the fact is, most PHDS probably don't have as many citations in their entire lifetime as Alec does in a single one.

Jason Wei

Jason Wei, an undergraduate at Dartmouth College, originally wanted to be a Wall Street banker, but he had trouble finding a finance internship during his freshman year (2017), so he was introduced to AI, and then it got out of hand.

As graduation approached, he applied for both a doctorate and a software engineer position. As for his PhD application, although he had a good conference paper, he was rejected by almost all schools except USC.

Because of his invitation to the Google AI Residency program, he chose to forgo studying for a PhD at USC and devote himself to AI research at Google.

In 2020-2023, he will work on Chain of Mind (CoT), instruction fine-tuning, and emergent phenomena in Google Brain.

In 2024, Jason Wei was mainly engaged in the development of o1 and o3, two models of thought chain reasoning through reinforcement learning, in OpenAI.

He has 48,338 Google Academic citations since 2020, and his most famous paper is "Chain-of-thought prompting elicits reasoning in large language models," published in 2022. The number of citations for a single paper has reached 10,914.

Greg Brockman

Greg Brockman is the co-founder of OpenAI and currently serves as its president.

After being admitted to Harvard in 2008, he stayed there for only a year before dropping out to attend MIT. He then dropped out of MIT after less than a year because college didn't give him the entrepreneurial environment he wanted.

So he went to Stripe to start his career and became their first CTO in 2013, growing the company from four to 250 people.

In 2015, he left Stripe to co-found OpenAI, where he served as CTO.

The degree is still the key but the ability is the treasure

For the above four special cases of OpenAI, the majority of netizens have also expressed their views.

Most netizens still feel that the reason why they don't need an advanced degree is because their own abilities are enough to produce influential work to prove themselves.

And most people still need a degree as a "key" to open many doors.

As netizen niv said, "If you already have a network and reputation, then of course, getting a PhD won't change much. But for the vast majority of people, these certificates can open many doors."

"Most people should still go to university, because people without a degree do face many challenges. The key is that people should know what they want and there are many alternative paths to achieve their goals," said Yuchen, who shared niv's view.

In short, the degree itself is a kind of "endorsement" of ability, after all, people who can get the corresponding degree, at least the learning ability will not be bad to where.

Therefore, the core lies in the ability itself, if you have enough ability to make influential research results, then the results themselves are naturally proof of your ability, and you do not need a degree to do your "endorsement".

But if you can't have the kind of research results that can prove yourself in a short time, then the degree can undoubtedly help us open a lot of doors.