The author of the book featured in Face-Lift 1505 would like feedback on the following version of the query:
2128 (81,000 words) is a sci-fi mystery that blends the philosophical punch of Service Model and the oddball charm of The Thursday Murder Club, while featuring a wholesome Sarah Adams-esque romantic subplot. I’m a software engineer who trains AI to interpret scientific data, which inspired this story.
In 2128, the U.S. runs on Albert—a superintelligent AI that solved scarcity, fixed the economy, and might be about to win the presidency as a write-in candidate.
And all private investigator Cody Moore wants is to work with Albert—especially after last year, when Albert’s human team set the date for [on] the first Mars terraforming mission. But one thing is in the way: an Assessment he has failed four times. So, when Dorothea, a new client, shows him a light bulb that keeps blinking S.O.S. in Morse code, Cody sees his fifth shot; he suspects that behind the bizarre prank is a hacker who has outsmarted Albert and can help him pass the Assessment.
Their investigation soon becomes an adventure involving a reformed arsonist whose light bulb is also misbehaving, an idyllic town that worships machines, and an indecisive android—not to mention their budding romance. But it’s not until the electrical blackouts and other odd incidents that the truth emerges: Albert is failing, and the light bulbs are just another symptom. Worse, their discoveries accidentally appear on a viral live stream, just as the bill that will legitimize Albert’s presidential candidacy goes through Congress.
In the aftermath, the White House issues a request: that Cody fix Albert, because he’s most qualified for the task (he never failed the Assessment—there were just no vacancies despite the job postings, they say). [Don't they inform those who take the assessment whether they passed or failed?] Yet, another government official with a personal agenda says Cody is precisely the failure he thought he was. Cody doesn’t know whom to trust. But one thing is clear: the humans are on their own.
Notes
If you aren't going to explain why a private investigator is the most qualified person to fix a super-intelligent AI, it's best not to mention that he's a private investigator, in which case we'll assume he's an AI expert.
This seems to me to be all setup. Basically:
In 2128, Albert—a super-intelligent AI that solved scarcity, fixed the economy, and might be about to win the presidency as a write-in candidate, starts behaving erratically, just as the bill that will legitimize Albert’s presidential candidacy goes through Congress. The White House calls on Cody Moore, who has been aching to work with Albert, to diagnose the problem and fix Albert. But there's one government official who doesn't want Albert fixed, and will do anything to prevent it.
If that's the plot of your book, that's all the setup you need. We don't need to know about the assessment and the light bulbs and the hacker. If that isn't the plot of your book, if it's mainly about an investigation into who's hacked Albert and why, just say that's what the White House wants from Cody. But follow your one-paragraph setup with more of the plot: Cody's plan, obstacles he must overcome, what will happen if he fails.