Book Review

Generative Artificial Intelligence, by Jerry Kaplan, Oxford University Press, 2024

One of the biggest challenges around GAI is keeping up with the field and separating out useful resources from noise. I put this book in the “useful” category and recommend it to everyone but existing experts. Dr. Kaplan provides and nice overview, circa 2023-24, with deeper dives into the underlying technology, GAI applications, and possible social impacts. This book will not answer every computer science question or data science question but does take steps in that direction. Dr. Kaplan has a PhD in Computer Science. For more, you will need to consult YouTube or local conferences. This field is moving at warp speed.
The book starts with a history, up to the first 1.5 billion neuron ChatGPT models in 2022. This is followed up by a tour of the recent technical milestones including transformer networks, encoders, decoders, tokens and emergent properties. The remaining seven chapters cover likely consequences of the AI revolution. Some highlights include legal (already happening); healthcare; education (already happening); software engineering (happening); creative arts (already happening). The recent hints by Sam Altman foreshadow a possible AI consumer device.
These applications will affect the workforce, which is Dr. Kaplan’s Chapter 4. Already we read about layoffs or potential layoffs attributed to AI. The conclusion is that the total number of jobs will not be reduced, but major transformations will occur. More and better retraining is one answer.
GAI risks are limitless and will be part of the development of the technology. The trick is not to focus everything on the fear of AI sentient beings. We still can wait for Ray Kurzweil’s “Singularity” in 2045, but not worry about it.
The most interesting chapter for me was on the Legal Status of GAI’s. This will be a fascinating area as GAI’s provide more agentic services for individuals and businesses. Key issues include property ownership by the GAI and liability in the event of disasters or accidents. The latter already is being litigated for self-driving cars. This chapter flows into the chapter on public policy (Chapter 7). Entrepreneurs might like to think we do not need regulation, and that technology is changing too fast to be regulated. The preceding six chapters would lead me to think that we do need regulation. In the EU, the AI Act was enacted in August 2024. China has a set of “Interim Measures’ focusing on their Socialist values. So far, the AI legislation in the US reminds me of privacy legislation. Everyone wants it, but no one is willing to put a stake in the ground.
Dr. Kaplan’s book is an excellent intellectual catalyst, encouraging readers to plan their own GAI journey.