In "ChatGPT and the Future of AI, " Terrence J. Sejnowski examines the transformative impact of artificial intelligence (AI) technologies like AlphaFold and ChatGPT, particularly in the realm of language models. AlphaFold, recognized with a Nobel prize, predicts protein structures using neural networks. Sejnowski's book, a follow-up to his 2018 work, delves into the evolution of AI from simple models to today's sophisticated large language models (LLMs) that mimic human language without true comprehension. Sejnowski highlights the architecture of neural networks, inspired by the human brain, and suggests that understanding their success could be as revolutionary as discovering DNA. He compares current debates about AI 'intelligence' to past discussions on the essence of life, hinting at future revelations that AI might bring about intelligence. The ultimate goal in AI is developing artificial general intelligence—machines that mimic human cognitive abilities.

Despite the current limitations, AI systems are becoming collaborators in fields like healthcare and education. Sejnowski emphasizes prompt engineering—crafting precise instructions—as key to effectively interacting with AI models. He suggests a "reverse Turing test" to assess the user's intelligence based on their interaction with AI. For future AI systems to achieve greater intelligence, they must mimic human childhood learning, integrating long-term memory, goal-oriented behavior, and sensorimotor capabilities. Sejnowski acknowledges the hurdles in AI’s path, including potential disruptions in industries and the need for eco-friendly models. Preparing for AI possibly surpassing human intelligence is crucial to mitigate risks. He advocates for ethical, well-regulated approaches to manage potential threats from AI, ensuring it benefits society.