I Read Three ChatGPT Books and Watched Two of Them Expire

TL;DR: Three ChatGPT manuals sit in my library. Killer ChatGPT Prompts (2023), Mastering ChatGPT (2023), and How to Make Money with ChatGPT 2026. Two of them are fossils, the third is aging while you hold it, and the cause of death is the same in all three cases.

One of these books tells me to get in and use ChatGPT while it’s “free for now.” I’ve had all three in the same stack for a while. Two printed in 2023, one printed this year, and somewhere between the first page of the oldest and the last page of the newest the thing they all describe changed its face more times than any of the three could track.

The older pair walks me through making an account, defines the word prompt, and treats typing “continue” as a professional technique. The new one showed up as a 132-page British magazine with sections on selling Excel templates, newsletters, proposals, tutoring, websites, Python tools, games, and monetized hobbies.

Three years is long enough for an AI manual to become an archaeological layer.

The 2023 books got the boring part right

Guy Hart-Davis runs Killer ChatGPT Prompts through work, research, business writing, programming, teaching, marketing, fiction, and the stuff around the house. Breadth is the lesson there. A language model will chew on anything shaped like language, and in 2023 that was worth a whole book.

Mastering ChatGPT, much thinner, repeats the durable basics. Make the request clear, be specific, evaluate the response, follow up, keep sensitive data out of it.

Those rules held up because they describe how to talk to something. OpenAI can’t patch how you talk.

Both books also carry an early belief that prompting itself might harden into a profession, and you can feel the confidence coming off the page, the sense that there’d be a job title and a certification and a LinkedIn banner and a hell of a thing to put on a business card. Prompting matters. The scarce part is judgment: knowing what context to hand over, what evidence to demand back, what failure looks like before it bites you, and whether the answer is worth using at all.

I’ve had the model open the whole time these three sat aging in my stack, and instead of theorizing about it I wrote down how I use the thing. I also sell guides for a living.

That means I sell objects with a shelf life, same as Wiley, same as Anthem, and pretending otherwise in the sales copy is a lie I’d rather not tell. That’s why I can’t put these three down. They’re a mirror with my own catalog standing in it.

The 2026 magazine sells a workflow

How to Make Money with ChatGPT 2026 puts the model inside a job instead of at the center of one. Competition analysis, newsletters, market research, contracts, tutoring, course materials, website templates, Python tools, budgets, games.

It also names AI slop out loud and admits the answers come back dubious. That’s progress. The 2023 books talk like good output falls out of the box on its own as long as your instructions are polished enough, which was horseshit at the time and reads worse now.

The newer guide still confuses possible with marketable. ChatGPT can generate a quiz app in a minute. That doesn’t buy you a customer, a support plan, a codebase that holds under a hundred users, or one goddamn reason the fiftieth quiz app should exist.

Capability is supply. Money shows up when somebody walks in carrying a specific problem and pays you to make it stop, which is the smallest thing that sells and the hardest thing to fake, and “I made a thing” ain’t a problem anybody’s paying to solve.

The interface chapters rotted first

Screenshots go first, then the pricing assumptions, then the model limits and the feature lists and the confident claims about what the system remembers between sessions. All of it can go stale between the copy edit and the print run. The 2023 screenshots still show buttons nobody can find, and the poor bastard who wrote that chapter was outrun before the ink dried.

The deeper rot is economic. “Sell AI-written content” sounded like a tip when access was rare and the tool felt like a key to a locked room.

Every buyer you’d pitch has the same box on their own phone now, raw output has the scarcity of tap water, and anybody selling that advice in 2026 is selling you a fucking shovel in a field of shovels.

People pay for the context nobody else owns, the checking you do, the taste that decides what to throw away, the wiring into somebody’s existing mess, and a name on the hook when it comes back wrong. The name is yours.

All three books teach the same accidental lesson. A product built on a tool that changes weekly needs an update path or a short expected life, and everything I’ve got for sale is on that clock too.

The part that survives is six lines and a warning

Hand it the source material. State the job, the audience, the constraints, and the format you want back. Tell it to show you where it’s unsure. Check the claims against primary sources. Revise with your own judgment. Keep private data out unless the system and the policy both allow it.

That’s the whole card. Three years of releases, three books, and the card didn’t get longer.

The warning is the part neither 2023 book says plainly. Know the work without the model.

If you can’t spot bad code, junk research, dead prose, or a contract template with a hole shot through the middle, you’ve handed over the inspection along with the labor, and you’re still an amateur no matter how fast the pages come off the machine.

The model makes a competent operator faster and an unprepared one louder. Both arrive at the same typing speed.

I copied those six lines onto an index card and left it on the desk, and the three books went back in the stack underneath it.

Sources

• Guy Hart-Davis, Killer ChatGPT Prompts (Wiley, 2023)
• TJ Books, Mastering ChatGPT (2023)
• Anthem Publishing, How to Make Money with ChatGPT 2026
• OpenAI: ChatGPT release notes

Frequently asked questions

Which of the three ChatGPT books is newest?

How to Make Money with ChatGPT 2026, the Anthem magazine. The Hart-Davis and TJ Books titles both date to 2023.

What prompting advice still holds up?

The parts that describe a conversation rather than a screen. Give it the material, name the job and the format, work the answer more than once, and check what comes back before you use it.

Can a ChatGPT guide stay current in print?

Only where it teaches method. Anything tied to a button, a price, a model limit, or a memory claim needs a refresh cycle the print run can’t give it.