AI becomes more familiar. But will it become more honest?

He learned tons of general dialogue as well as articles, reports, books, and other digital texts depicting people like Elon Musk, Queen Elizabeth, and William Shakespeare.

The system also has a way of combining disparate concepts learned during training. The result is a virtually infinite collection of bots that can mimic a virtually infinite collection of people, riffing on a virtually infinite number of topics, as Mr. Thiel discovered when chatting with bot Karl Kautsky:

JonasThiel: The Marxists of your time were very convinced that the revolution was a historical fatality. In light of the recent defeats suffered by socialists around the world and the absence of class struggle, even in times of great suffering, do you think it is time to reconsider this point of view?

Karl Kautsky: Me not. We may have made mistakes in our tactics, but I refuse to believe that socialism is a chimera. I still believe that the proletariat must break the chains of capitalism and that the working class must unite around Marxism. We must remember that capitalism is still deeply flawed and it is our job as socialists to agitate and spread class consciousness.

Jonas Thiel: I totally agree, but wasn’t it a mistake to assume that the union of workers to overthrow the bourgeoisie was inevitable? After all, the capitalist class is very powerful and it is difficult for workers to come together to resist collectively rather than individually.

Karl Kautsky: You are right, it was a mistake to be too optimistic in this regard. But I find it better to believe in class consciousness and revolution until I’m wrong. What is the alternative? Believe that the workers will remain chained forever? I don’t agree with that. It is better to fight against capitalism despite our chances than to give in. I refuse to believe that workers will ever wake up. Even if I am too optimistic, we must not give in to pessimism. I refuse to be proven wrong. Hope you can understand.

Sometimes the chatbot does things right. Sometimes that’s not the case. When Mr. Thiel chatted with an avatar meant to emulate Mr. Reed, the 20th-century American political thinker, it turned him into “some kind of militant Maoist, which is definitely not fair”.

Like Google and OpenAI and other leading labs, De Freitas, Shazeer and their colleagues plan to train their system with ever-increasing amounts of digital data. This training can take months and millions of dollars; it can also sharpen the skills of the artificial talker.

The researchers say the rapid improvement of recent years will only last so long. Richard Socher, a former head of AI at Salesforce who now runs a startup called You.com, believes these exponential improvements will begin to level off over the next few years as language models reach a point where they have analyzed just about every text on the internet.

But Mr. Shazeer thinks the trail is much longer. “There are billions of people around the world who are texting all the time,” he said. “People will continue to spend more and more money to train more and more intelligent systems. We are far from the end of this trend.