The game of Language

ChatGPT is surprisingly smart and yet remarkably stupid. But same as humans, many academically good students are able to solve difficult math or physics problems, but don’t exhibit any sign of critical thinking or even intuitive understanding. The boundary between a chatbot and a real individual is becoming harder and harder to identify. AI and science right now are accelerating at an exponential rate, while for the human brain, the rate is linear. Making the result of a million years of evolution might be caught up by silicon in the next few years. One of my favourite scientists, John von Neumann, coined the term “technological singularity” when AI surpasses human intelligence, then overthrows homo sapiens’ no.1 position in the ecosystem. I hope by that time, human civilization lives like the Dune plot and witnesses Lisan Al Ghaib.

Coming back to distinguishing humans and ChatGPT, we have to derive the solution from the root of language, which is used to massively train ChatGPT. Another favourite philosopher, Ludwig Wittgenstein, thinks of language as a multiplicity of language games, and we play the games every day, and the beauty of it IMO is that we’re playing unconsciously. A normal conversation is a communication game when players exchange stories and ideas that bring bonding rewards to both. The language game can advance sophisticatedly into fraud, gaslighting, or even brainwashing. Thinking about fraud, one person’s incentive is money, and they intricately create a convincing plot to gain trust from the victim. It’s amazing to me how they improvise, which keeps the game unrecognizable to the victim; the sense of playing in words is unfathomable. And well, the same principle applies to gaslighting and brainwashing. I personally enjoy the moment when I take a course of sentences, and the opponent responds in a predictive manner, or sometimes I unravel the deep ego of individuals. Be aware of what games you are playing, or else you lose without knowing that you lose. Some big winners win the whole country with words and myths.

That’s the game of language. A game must consist of players, rules, and rewards (the most important factor). ChatGPT’s reward was aligned with how humans speak, how humans write, but not how humans play the game, and no amount of text can reveal the hidden intentions, driving the improvisation. The lack of true understanding of language games explains ChatGPT’s smart-stupid duality, acting as an “imposter intelligence”. That’s the core difference for humans, as we all have the incentives to do something, to say something. When an AI system capable of conducting “đa cấp” over the elderly, that’s the end for us.

As a student, trying to push the limits of AI in different aspects, it’s hard to not think of the undeniable fact that AI is taking over jobs, but AI has no intention of doing that. The intention must come from somewhere else, and the answer is the capitalist incentivizing AI. So the next time anybody worries about losing jobs to AI, the right question to ask is: “Do people want this job to be replaced by AI?”

P/s: Wittgenstein’s core idea is that a single word can be interpreted in different meanings. The improvisation in meaning’s space depends on the context or the game.