Don't Be Fooled by the Next AI Scam
Wolfgang von Kempelen's mechanical wonder stumped audiences for 50 years...
His "Automaton Chess Player" was a machine disguised as a life-size moving doll dressed in Turkish clothing. And as the name suggests, it played a mean game of chess...
The chess player eventually became known as the "Mechanical Turk." It moved its own pieces, executed strategies, and won – a lot. It even beat opponents like Napoleon Bonaparte and Benjamin Franklin.
Kempelen claimed to have built the first machine that could think on its own. But it was all a hoax... A human chess master hidden inside controlled its motions in secret.
Today, we still use the phrase "mechanical Turk." It describes situations where human agents are posing as "smart" machines.
This month, a billion-dollar AI startup was exposed as perhaps the biggest mechanical Turk in history. And it serves as a major warning for AI investors...
I'm talking about an AI startup called Builder.ai.
Founded in 2016, the company's mission was to make building an app "as easy as ordering a pizza." It was supposed to bring software development to the masses.
Users would share their ideas with a chatbot named "Natasha" on the Builder.ai platform. Natasha would then give them detailed feedback on how to build the app.
Builder.ai grew to a valuation of $1.5 billion. It even attracted Microsoft's backing in 2023...
But last month, Builder.ai collapsed. It turns out, Natasha was actually 700 Indian engineers writing code manually.
The massive 10-digit scale of this fraud should give investors pause. Builder.ai fleeced investors out of millions of dollars. It even fooled Microsoft – one of the best businesses in the world. All it took was futuristic promises and a glitzy package.
As the world continues to adopt AI, many other small, overpromising AI companies will emerge.
Buying these smaller-scale businesses will seem like a great way to multiply your investment gains in AI. But many of them will be mechanical Turks. Still more will be "wrappers" – an interactive interface slapped onto an existing AI model like a fresh coat of paint, rather than a true startup's new model.
How to Win the AI Race
For a real edge in AI, companies need to build proprietary models of their own. But that takes tremendous resources, including...
- Energy – chatbots currently require up to 1,287 megawatt-hours of electricity per training run. That's roughly what it takes to power more than 100 U.S. homes in a year.
- Money – the cost of training large language models is increasing exponentially. For example, OpenAI's GPT-2 cost $40,000 to train. But the cost soared to $100 million for GPT-3.
- Data – chatbot training requires vast amounts of unique text. For example, the entirety of English-language Wikipedia contains around 4.6 billion words. GPT-4 took 9.75 trillion words to train. So GPT-4's training data contains more than 2,000 times more text than Wikipedia.
These three things are essential to build a competitive AI model. Any company without them will fall behind in the AI "gold rush" – no matter what claims it makes.
Builder.ai is the biggest AI company to flop since the AI race took off in 2022. But it won't be the last. Many more companies will make untrue claims about their AI capabilities in the coming years...
So be cautious of the AI "little guys" that promise too much. To safely invest in AI, buy the true hyperscalers: the companies that have the resources to build their own models. Meta Platforms, Alphabet, and Amazon are all hyperscaling AI plays.
Otherwise, you can buy the "picks and shovels" of AI... semiconductors. The VanEck Semiconductor Fund (SMH) is a great way to buy all the best semiconductor companies with a single click.
However, I would resist the urge to cherry-pick from the army of AI startups that's on the way. Many investors will be fooled by mechanical Turks in the coming years – but you don't have to be one of them.
Good investing,
Sean Michael Cummings
Further Reading
"For the first time in years, Alphabet is quietly winning the AI arms race," Sean writes. After fumbling its first attempt at building an AI chatbot, Alphabet has transformed itself into an innovative AI platform. But investors are missing how cheap its shares are today.
"Don't fall into the 'I missed it' trap," Whitney Tilson says. Value investors tend to only look in the bargain bin for beaten-up stocks. But that puts them at risk of falling for value traps – and missing out on some stellar winners.
HIGHS AND LOWS
NEW HIGHS OF NOTE LAST WEEK
Nasdaq (NDAQ)... stock exchanges
Barclays (BCS)... financial services
Visa (V)... credit cards
Automatic Data Processing (ADP)... payroll giant
Microsoft (MSFT)... tech giant
IBM (IBM)... computers
eBay (EBAY)... online marketplace
Carvana (CVNA)... online used cars
DoorDash (DASH)... food-delivery app
Chewy (CHWY)... pet products
Cheesecake Factory (CAKE)... restaurants
Cintas (CTAS)... uniforms
Johnson Controls (JCI)... industrial equipment
Heico (HEI)... aerospace and defense
New Gold (NGD)... gold
NEW LOWS OF NOTE LAST WEEK
Not many... It's a bull market, you know!
Join for free. See how over a million people are taking control of their wealth…
