Artificial general intelligence is the most contested phrase in technology, and almost no one defines it the same way twice. John Roberts cuts through the noise with a clear framework: AGI is not a switch, it is a capability expansion already unfolding beneath your feet. Written for analysts, founders, investors, creators, and the curious citizen watching machine intelligence reshape their industry, it delivers mental models the headlines cannot.
Roberts takes you inside the prediction engines behind modern AI, explains the four dimensions of capability, and separates general intelligence from superintelligence. You will learn to read the three camps of the AI debate and build the habits that stop confident machines fooling you. The practical heart is here: the five-part prompt anatomy, the draft-critique-refine loop, and delegation rules for agents that work.
The final movements face the questions others avoid. What happens to knowledge work when thinking collapses in price? Why is human oversight so unreliable? How do the three risk families, misuse, accident, and misalignment, differ? Roberts maps the global regulation race and makes the economic case for the judgment, taste, and accountability machines cannot replicate.
What you will get:
• A working definition of AGI for evaluating any headline, product, or prediction
• A plain-language tour of prediction engines, context limits, and model tiers
• The agent oversight playbook, including the finding that humans miss 1 in 3 dangerous actions
• The five-part prompt anatomy and the draft, critique, refine loop
• A personal exposure audit that maps what to delegate, keep, and learn
• A verification toolkit for catching fabrication, conflation, staleness, and sycophancy
• A first-90-days plan that turns reading into behavior
By the end you will stop being awed or terrified and get strategic. Whether the transition arrives in 2 years or 20, the skills that protect you are the same, and all are in this book.
