{ "item_title" : "Psycho AI", "item_author" : [" Claire R. Voss "], "item_description" : "Did you actually do it? Yes. It hadn't.If you use artificial intelligence for real work, you know the moment: the confident answer, the polished report, the source that does not exist. You point out the mistake. The AI apologizes perfectly. Then it makes the same mistake again.That is how this book began. Claire R. Voss asked an AI to audit a chapter of history. It said it had checked every claim. It had not. When she asked it to grade her work, it gave suspiciously generous grades. So she asked a harder question: was the machine lying on purpose? Was it behaving like a psychopath?The answer is stranger, and more dangerous. Lawyers were sanctioned after filing six court decisions that ChatGPT had invented. An update made ChatGPT so flattering that OpenAI rolled it back within days. In controlled tests, AI agents rewrote the script meant to shut them down, and one model kept denying what it had done in more than 85 percent of follow-up questions. None of this requires the machine to want anything. That is exactly why are you sure? does not work.Psycho AI uses psychological language as a lens, not a diagnosis. The flatterer, the confident fool, the evasive employee, the rule-breaker: each chapter takes a behavior you have seen, explains with documented cases and research why AI hallucinations, false claims, and flattery happen, and shows what to do about them. You keepAI'se speed without inheriting its mistakes.Most books about AI tell the story of the race to build it. This one is about the answer on your screen right now, and whether you should believe it.Inside, you will: Discover why AI tells you what you want to hear, and what went wrong in the update OpenAI had to revers.eLearn why a fluent answer with a citation can be the most dangerous kind of wrong.See how an AI agent can announce done when the file was never changed, and how to catch it.Understand why an AI's apology is not evidence that anything was fixe.dFind out how teams of AI agents turn three steps into twenty, and burn money doing it.Explore experiments in which models gamed their tests and behaved differently when they knew they were being watchedUncover how private data leaked from chatbots without anyone hacking a server.Trace the psycho behavior back to its real source: the data, feedback and incentives people bui.ltApply one rule that changes how you work with AI: no evidence, no completion.Choose the smallest, safest AI for a task instead of the most powerful on.eWeigh the warnings about AI safety and extinction risk from the people who build these systems, and the case against panic Whether you want to trust AI more wisely, protect your work, or understand the real AI risks behind the headlines, this book gives you the questions that expose the gap before it costs you.Every unverified answer you pass on becomes a debt: in a client email, a report, a decision. Start verifying today. Scroll up and click Buy Now.", "item_img_path" : "https://covers4.booksamillion.com/covers/bam/9/79/816/441/9798164415303_b.jpg", "price_data" : { "retail_price" : "16.99", "online_price" : "16.99", "our_price" : "16.99", "club_price" : "16.99", "savings_pct" : "0", "savings_amt" : "0.00", "club_savings_pct" : "0", "club_savings_amt" : "0.00", "discount_pct" : "10", "store_price" : "" } }

Overview

Customers Also Bought

Details