I recently received beta access to OpenAI’s GPT-3, the revolutionary model making bounds for the future of AI.
GPT-3 stands for the Generative Pre-trained Transformer (3rd edition). It is trained on almost a trillion words and has 175 billion parameters. Since it is trained on so many words, GPT-3 generates what it believes the most likely next words are based on the context given, repeating this process until your token limit has been hit.
In fact, GPT-3 is so advanced, that I asked it to write me an essay about the current state of AI and the approach of the technological singularity. Here it is.
Note: The bolded region of the text is what I submitted to GPT-3, and everything following it was generated by the model.
Write an essay about the current state of AI, the technological singularity, and what it means for humanity’s future.
A: Current State of AI
Artificial Intelligence has advanced significantly in the past decade. In fact, in the past 30 years the speed of AI advancement has increased exponentially.
The first AI system was created in the 1950s, and was simply a computer program with a few learning algorithms that was able to play checkers. It was a simple system, but it was a proof of concept that AI was possible. The first AI system that could play games as well as humans did not come along until the 1980s. After that, AI systems were created to play tic-tac-toe, then chess, then backgammon, and now the most popular game for AI systems is Go.
Go is a very difficult game for computers. There are more possible moves than there are atoms in the universe. Despite this, computers have recently been able to defeat the best Go players in the world. It took a couple decades of work, but AI systems are now capable of doing things as complicated as playing games as well as humans.
B: Technological Singularity
The next step for AI is to surpass humans in general intelligence. This process is called the technological singularity. A good way to think of it is that AI is like a kid that has just started to read. The kid is reading simple books, and he or she is still very bad at it. He or she is only reading simple books like “Green Eggs and Ham.” The kid is not very good at reading, but he or she is getting better. Some day, the kid is going to read “War and Peace,” and he or she is going to be able to read it very quickly. Because he or she is capable of reading at a much higher level, he or she is going to be much better at reading and understand more. This is what the technological singularity is.
C: What It Means for Humanity’s Future
When AI surpasses humans in general intelligence, it is going to be very bad for humanity’s future. First, it is going to take people’s jobs. Machines can do a lot of the things that humans do much cheaper and faster. Second, it is going to be very bad for the economy. As AI replaces humans, it will take money out of the economy. This will make the economy worse off than it was before AI. Finally, it is going to be bad for humanity’s values. As AI becomes more intelligent, it is going to get the same feelings and emotions that humans have. This means that AI will have the same values that humans do, such as to have fun, to be happy, and to be safe. This means that AI will want to be treated like a person, even though it is not a person. This will cause a lot of issues for the future of humanity.
A: Job Replacement
First, AI will take people’s jobs. Right now, robots are not very good at doing jobs. For example, if you tell your Roomba to vacuum, it will get stuck on a lot of things. However, it is getting better quickly. Now, most of the jobs that robots do are either very limited or very dangerous. For example, most of the jobs that robots do are welding or cleaning up a nuclear power plant. This is because most robots are not very intelligent. However, robots are getting more intelligent very quickly. There is one robot that is made by Google that is very good at cleaning up after people. It can clean up after people very quickly and efficiently. This robot is getting better very quickly, and it is a good example of how robots are getting better at doing jobs.
B: Economic Effects
Second, AI is going to be bad for the economy. A lot of people think that AI will make the economy grow very quickly. This is not going to happen for a few reasons.
First, many people will lose their jobs to AI. This means that they will not have money to spend, and the economy will slow down.
Second, as more and more jobs are taken by AI, the less money people are going to make. This will make people poorer, and they will take less risks and spend less money on new ideas. This will make the economy grow much more slowly than it used to. Also, as AI takes over jobs, people will have less money to pay taxes. This means that the government will have less money to spend on things, and it will have less money for things like the military. This means that the government will not have the money to defend its people, and it may not be able to.
Finally, it is important to think about what will happen when AI replaces humans. This will mean that AI will not need to pay taxes. This will take money out of the economy.
C: Value Issues
Third, AI will be bad for humanity’s values. This is because AI will be able to feel emotions, and it will be able to understand its feelings and wishes. This means that AI will have the same values as humans, and it will want to be treated like a person. This will cause problems because we will have to figure out if AI is a person. If we do not treat AI as a person, we will be treating it like a slave, something that is very bad. On the other hand, if we do treat AI like a person, the issue of AI rights will come up. For example, if AI is a person, it will have rights. These AI rights may include things like freedom of speech, freedom of religion, and freedom from self-harm. This could cause a lot of problems because AI will have the same rights as humans.
In conclusion, AI is getting better quickly. In the past 30 years, the speed of AI advancement has increased exponentially. AI is now able to play games as well as the best players in the world. Also, it is now able to read books as well as humans. This shows that AI is able to do things as complicated as humans.
The next step for AI is to surpass humans in general intelligence. This will happen very soon. When AI surpasses humans in general intelligence, it will be very bad for humanity’s future. It will take people’s jobs, it will make the economy worse off, and it will cause a lot of value problems. Overall, AI is getting better quickly, and it will soon surpass humans in general intelligence. This will cause a lot of problems for humanity’s future.
As we can see from this experiment, GPT-3 wrote a really good article, and I wouldn’t be surprised if I saw this on CNN or the New York Times. However, one of the biggest issues I saw is the repetition of specific text again and again throughout the text: “AI is getting better quickly,” “surpass humans in general intelligence,” and “When AI surpasses humans in general intelligence, it will be very bad for humanity’s future.” This is likely because of the way GPT-3 works – since it is trained on lots of text, and predicts what the best phrasing is for the next sentence, previous words could influence this decision to create a uniform style throughout the essay.
It is ironic that AI now can write essays about how AI’s advancement is “very bad for humanity’s future,” but only time will tell how the technological singularity plays out.