The "Ponzi Scheme" of Current AI

A PhD in CS/AI is no longer what I understood it to be.

The "AI for Science" Phenomenon in Academia

Go look at almost any PI's website outside of the "top labs" on CSRankings. Everyone is doing healthcare, biology, chemistry, or medicine. It is not only happening in unranked or R2 schools; it is happening everywhere outside of the top labs. More than half of the professors are in this field, and I think they are doing it not because they like it, but mostly out of desperation to get NSF funds. Even if you look at "networks" or "database" professors, a lot of them are running to it as well. Many professors themselves were originally in databases or OS; then they shifted to NLP as the GPT hype rose, and now they have pivoted to healthcare because they need grants and cannot compete with industry compute. In fact, if you search for recent grants, they are all about this shit.

Most papers here are frauds with no reproducibility. It doesn't matter if you are at an R2 school, a 150 ranked school, or at a top 20 school doing healthcare; it is probably academic fraud anyway. If you write papers, you need to read papers. If you read papers, you need to know what they are talking about. If you try to understand what they are talking about, you can't. If you try to email the authors, you get no replies, no data, and no source code. Ultimately, I do not actually want to do applied research.

The Reality of a Green Card/"Research Scientist" ("Why do a PhD at all?")

What do I do if I am a mid-GPA student at a school like USTC, who is good at math compared to the general public but has no papers? I am not talented enough to be a literal genius, guarantee to have profound and significant breakthroughs, or lead open-source projects right now. Furthermore, I do not have the financial cushion or network to launch a startup directly.

My goal is to go to the US or a Western country to become a research scientist. If I go to an MS program, it is usually a huge factory churning out a thousand students. I do not want to be a Software Engineer (SWE); there are too many of them, and it is mostly an applied role. So my options feel limited. I can't get into the top 4 PhD programs to achieve this goal. So the best way for me is to get into a mid-tier PhD.

The Speed of Research

To be a researcher in a top lab, you need to go to a top 5 PhD program. To get into a top 5 PhD program, you need to have multiple papers in undergrad. To join labs in undergrad, you need to handle the courses well and have abundant free time. I do not fit this profile from the start, but I do not think I am denied entirely and should become nihilistic.

The LLM hype is literally younger than my backup laptop. By the time I finish my PhD, nobody will care about what I did during my doctorate, because the technology will be outdated in a few years anyway. Since PIs don't actually know what they are doing now, their guidance might not matter that much anyway.

Chill

Find a chill tenured professor. If you can't work for the professor's grant, then be a TA and compromise if possible. Forgive the people who are committing fraud. Avoid resentment at all costs.

If you are in a state of terminal anxiety or terror, you do not do good research, and progress grinds to a halt. True research is built on curiosity and passion, not on terror and anxiety. Therefore, you should not sacrifice your well-being for marginal research, a marginal GPA increase, or marginal anxiety and terror. Do not be scared of other people churning out papers or doing whatever else. Do not panic when someone else beats you. Remember: capitalism is essentially broken, detached from the human soul, and not a true reflection of human value. Even if you aren't making as much as others do, an AI researcher in the current economy is already running circles around most professions. Have a low ego, but maintain your dignity. Ultimately, starting in my early 20s, I want to build a 50+ year career.

What to Optimize For? And Where Do I Go in the Future?

Health (Mental Health, Physical Health)

Have fun, eat good food, stop being anxious and sleep well, enjoy my youth, participate in sports and retreats, and hang out with beautiful people.

The great delayed gratification lie I was sold on the first day of middle school has turned into perpetual delayed gratification. They say middle school is busy, so you have to study now. Then high school. Then university. Then a PhD. Then a corporate grind. In January you have finals. In February you have projects. In March you are busy. In April you are writing a thesis. In May you have midterms. In June you have finals. This lie must end. There is nothing at the end of it but nihilism and an existential crisis. Since I already lost so much time to middle school, high school, and my undergrad, there is no way to pay it back. But now, after grieving for those lost teen years, I need to make sure it does not happen again.

Communication Skills

An awkward, introverted person gets heavily penalized in the modern capitalist economy. Therefore, you must be extroverted, proactive, and advocating for yourself.

Also learn how to present and communicate with people outside of your subject. Do not look down on or resent people outside of the field, and do not harbor an elitist complex. Be friends and have empathy for non-tech people as well. Start a social media channel and organize social events. This isn't endless partying, drinking, or chaotic behavior. This is simply about learning how to network, market yourself, and communicate effectively.

Future

A PhD opens the door for me to apply to PostDocs or "Research Scientists" in better universities. As I learned, a PostDoc is much easier to enter than a PhD and it is great for networking. An unpaid Research Scientist at MIT can have very few requirements and it can be a brilliant PR (such as Lex Fridman). I have zero intentions of entering the Academia.

I do not necessarily need to work at Anthropic or Google. There are many lesser-known labs (like AI21, Cohere, or Nebius), or I could also just join a startup. If I have some money and live comfortably, I can become a founder or pivot to any field I like. Only after the PhD do you truly hit the road.

TLDR

• Play the basic conventional game: Get the "Doctor in Computer Science," secure a Green Card, and build solid basic coding skills
• Prioritize health and socializing: Protect your physical and mental health, develop soft skills, and never allow yourself to become isolated, socially stunted, or delay dating.
• Use these two foundations to aim toward a "Research Scientist" job. Once the car hits the road, you can do whatever you want.