"No company is immune to the AI ​​bubble." —Sundhir Pichai, Google CEO
"We're in an AI bubble right now." —Sam Altman, CEO of OpenAI

★ “If you haven’t read this book, don’t invest a single penny in AI stocks!”
★The first "AI bubble" detection guide written by a computer scientist named one of Time's "100 Most Influential People in AI."

AI will wipe out humanity? Wrong.
The real danger is 'fake AI' targeting your wallet.
The two authors, computer scientists at Princeton University and at the forefront of AI research, assert:
“We are buying a fantasy, not a technology.” This book dispels the exaggerated marketing of “generative AI,” represented by ChatGPT, and scientifically proves how “predictive AI,” which determines our lives in areas such as employment, healthcare, and public safety, operates like a 19th-century “snake oil” scam.
This book presents the most sobering and practical criteria for distinguishing between "real innovation" and "fake technology" in the trillion-dollar AI market.

Imagine a world where there is only the collective noun 'vehicle' and no separate words for the various modes of transportation.
The people who live there rely solely on the word "vehicle" to refer to cars, buses, bicycles, spaceships, and any other means of transportation.
Conversation in such a world is bound to be confusing.
Some people think of bicycles, others of trucks, and there are heated debates about whether or not these 'vehicles' are environmentally friendly.
Now, let's take a closer look at the world we live in by replacing the word "vehicle" with "artificial intelligence."
--- p.17~18, from “Chapter 1: Facing the AI ​​Landscape”

AI used in recruitment is a bleak example.
In the United States, three-quarters of employers use automated tools to screen job applicants.
Some tools screen applicants based on their resumes, while others test applicants through video interviews.
These tools are used first in the applicant screening stage.
If a person fails the entrance exam, his or her resume is discarded without a single human glance.
But the process is unclear.
Companies don't share how their recruiting software is built, leaving applicants in the dark about how they're being judged.
--- p.78, from “Chapter 2: How Predictive AI Fails”

While the existing limitations of prediction can be overcome in some cases by improving the quantity and quality of data, they also exhibit inherent limitations.
In some cases, such as predicting which cultural products will do well, we never expect the predictability to improve much.
While predicting individual lives may improve somewhat, we cannot expect drastic changes.
Unfortunately, that doesn't stop companies from selling AI as a predictive tool for making important decisions about people.
Therefore, rather than passively hoping for improvements in predictive AI technology, it is important to resist the snake oil AI already widely used today.
--- p.147, from “Chapter 3: Why Can’t AI Predict the Future?”

A New York attorney wrote his brief with the help of ChatGPT.
It seems they forgot to include a disclaimer that this chatbot may generate inaccurate information.
The statement of purpose was full of cases that did not even have precedents.
The lawyer asked the chatbot if the incidents were real, and the chatbot answered yes.
Without even being able to recognize that the events in question were fake.
Even the overall verdict was manipulated.
The lawyer submitted a document based on the chatbot's answers, and, as expected, the presiding judge was left in a predicament. The sight of lawyers being penalized for submitting sloppy documents using AI has become a common sight.
--- p.154~155, from “Chapter 4: The Long Journey to Generative AI”

In 2023, the Future of Life Institute published an open letter urging AI labs worldwide to immediately stop training AI systems more powerful than GPT-4, even for just six months.
Simply put, many in the AI ​​community argue that AGI poses an imminent existential threat that requires bold global action to prevent.
If this is true, then nothing else in this book is important, and there is nothing more important in the world.
In the following, we will reveal why these claims are based on a foundation of unfounded wisdom. This is not to say that AGI will never be built, or that even if it were, there's nothing to worry about. AGI is entirely feasible in the distant future, and society already possesses the tools to calmly address the risks.
--- p.219~220, from “Chapter 5: Is Advanced AI an Existential Threat?”

In 2016, Facebook sparked widespread outrage by deleting a photo commonly known as the "Napalm Girl" as soon as it was posted.
The photo in question shows a nine-year-old girl named Phan Thi Kim Phuc running naked in agony, badly burned by a napalm attack.
It played a historically significant role in publicizing the horrors of the Vietnam War and changing public opinion about the war, and is considered one of the most iconic photographs in the world.
At first glance, this photo blocking incident seems like a good example of the limitations of AI we've been talking about.
--- p.288~289, from “Chapter 6: Why Can’t AI Fix Social Media?”

For example, OpenAI claimed that “ChatGPT-4 achieves human-level performance on most professional and academic exams,” and indeed, ChatGPT-4 scored in the top 10 percent on the bar exam.43 Many took this as a sign that AI would soon become good enough to replace lawyers.
But a lawyer's job isn't all about answering bar exam questions all day long.
Real-world usefulness is different from good performance in benchmarks.
Moreover, professional exams, especially the bar exam, tend to overemphasize the ability to memorize knowledge and underestimate practical skills, which are much more difficult to measure with standardized tests.
So, not only do these tests fail to capture the true utility of AI, they also only highlight what AI does well.
--- p.336~337, from “Chapter 7: Why Do Myths About AI Persist So Tenaciously?”

Today, companies spend tens of millions of dollars on advertising to avoid regulations.
When antitrust legislation was introduced to prevent Big Tech companies from favoring their products over those of their competitors, advocacy groups funded by the companies poured $36 million into advertising opposing the bill.
Supporters of the bill, who did not receive funding from these companies, spent $200,000 on pro-regulation ads.
Compared to the former, this was an amount almost 200 times less.
Big business has also funded advocacy groups to make it appear that small businesses are opposing regulations.
For example, a group called American Edge, backed by Facebook, has been using small business owners across the country to voice their opposition to regulations and has run ads that stoke fears that American businesses might be pushed out by China.

--- p.381~382, from “Chapter 8 │ Where Are We Going From Here?”