The False Promises of AI
How tech companies are fooling us
In 1770, the Hungarian author and inventor Wolfgang von Kempelen introduced an automaton chess machine known as The Mechanical Turk. This device showcased its automated chess master skills across Europe, frequently emerging victorious in matches against human opponents. It even reputedly defeated notable figures such as Napoleon and Benjamin Franklin. The Mechanical Turk rapidly gained immense popularity and was hailed as a marvel of its era. However, the excitement surrounding it eventually unraveled as a deceit concerning its "autonomy" when it was revealed that a person, hidden beneath the table, was actually controlling the machine. This person covertly orchestrated the game's strategies from that hidden position.
In simpler terms, it was a profound deception that everyone believed at the time.
Nearly 250 years later, in 2016, Amazon executed a similar stunt. Its "Just Walk Out" payment system enabled customers to pick up items and exit without manually scanning them, creating the illusion of transactions and logistics being autonomously managed. However, the real breakthrough in this AI advancement, which integrated technologies such as Computer Vision, Sensor Fusion, and Deep Learning, relied significantly on about 1000 individuals in India. These personnel monitored operations and ensured the precision of every payment.
Currently, we see that it's often humans who undertake the task of labeling data, since this is how AI models are initially trained. This approach is necessary and appropriate. However, the issue is that the way AI was initially marketed to us was different and misleading.
In 2022, these 1,000 individuals were still manually reviewing 70% of the transactions in 20 Amazon GO stores, 40 Amazon Fresh grocery stores, and 2 Whole Foods Stores.
While some may see it as somewhat dystopian, Amazon touted its technology as a magical, AI-driven solution.
The real problem is that companies like Amazon, along with many other major firms, are not fully transparent about how these significant AI-related advancements actually operate. Given the AI revolution, it has become clear that we need a more critical examination of what’s really happening.
AI-Washing
It’s not just you and me noticing that the term AI is everywhere these days, far more than in previous years. This technology has shifted from being merely a trendy topic to a daily discussion point. Until before 2022, the term AI was mostly confined to research articles, which didn't generally catch the public’s interest. Even the launch of GPT-3 followed this pattern. However, everything changed post-early 2022, especially after the launch of ChatGPT. Suddenly, social media and websites were flooded with AI-related news, and terms like “powered by AI” became commonplace (though, in my opinion, somewhat overused). While the widespread use of these terms is justified in many cases, they all exhibited a common issue known as AI-washing.
Simply put, AI-washing occurs when companies generate false hype and deceive investors by sharing misleading information about the capabilities and risks of their AI products, or by lying about how and when they use AI.
Can you recognize any such cases?
According to Goldman Sachs, 36% of S&P 500 companies mentioned AI in their fourth-quarter earnings reports. If the world's largest companies are openly promoting this technology, it’s clear that smaller companies are too, yet many without tangible results to support their claims.
Global corporate investments in AI have increased sevenfold since 2015. Many companies are experiencing significant growth in the recent AI surge, which pressures other companies to start integrating AI into their business models or products to stay competitive. But are the outcomes really living up to the hype?
The Canadian investment firm Delphia claimed to have developed an AI capable of predicting the next big companies and industry trends. However, following an investigation by the SEC, it was found to be a fraud, and the AI product did not possess the capabilities claimed. Consequently, Delphia was fined $225,000. Another example involves the CEO of Wirecard, Markus Braun, who boasted of having patented AI technology for all his fintech products. In reality, no such advanced technology existed; the operations were merely conducted on spreadsheets.
It's highly likely that you've heard about what a Neural Processing Unit (NPU) is at some point. This is a processor specifically engineered to handle tasks with the assistance of AI. As you might imagine, PCs and laptops are incorporating these new tools to offer users a unique experience with this technology. However, many users have expressed dissatisfaction with the quality of the generated responses in recent product reviews. Essentially, they find them unusable. Chris Hoffman sums it up by saying, "aren't all they are cracked up to be… and if you’re expecting something transformative when you buy one at the start of 2024, you’re going to be disappointed… They might one day deliver a lot of cool features—just not yet."
At this juncture, you might be wondering if there are any consequences to such misplaced promises other than eroding customer trust in companies?
With the rollout of AI products, there is a constant push to dazzle us and uphold this false expectation or facade. Not even the world's most renowned companies are immune to falling for this ploy. Arguably, the most significant consequence of AI-washing is that it makes us susceptible to any novelty presented as an "opportunity"
AI Bubble
Almost as soon as AI began gaining significant popularity towards the end of 2022, many people immediately started drawing parallels to the internet bubble or the cryptocurrency craze. In fact, a sizable group still views it this way. If we examine the internet or “.com” bubble more closely, the issue wasn’t the World Wide Web itself, but the e-commerce aspect, which promised to attract hundreds of investors. This did not materialize to the extent anticipated, leading to a crash when it became clear that the companies they had invested in weren't profitable.
Now, concerning AI, there's still a cautious sentiment among investors, which means that the growth we are currently experiencing is moderate.
Peter Oppenheimer, Chief Global Equity Strategist at Goldman Sachs Research, states: “We believe we are still in the relatively early stages of a new technology cycle that is likely to lead to further outperformance.”
Additionally, it’s noteworthy that NVIDIA is designing chips that are fueling the AI revolution among tech giants. This is evident in the performance of their stock, which has seen an 80% increase just this year, 2024. While it might seem excessive, it reflects the market's perception of AI.
Mark Cuban, who made his fortune during the internet boom, also does not see AI as a bubble. In a recent interview with Lex Fridman, he noted that the most significant indicator that we are not in a bubble is the lack of Initial Public Offerings (IPOs) in the AI sector. The absence of overvalued companies trading on the stock market and the scarcity of AI company listings are key indicators. Moreover, Cuban points out that the current market does not display these characteristics.
There’s a recognized pattern for emerging technologies known as the Gartner Hype Cycle. It describes how humans tend to get overly enthusiastic about a new, revolutionary technology, overestimating and magnifying its impact. After this phase of hype, the market naturally collapses. Subsequently, the companies that survive re-enter the market with genuine value and lead the development of the new technology until it matures.
If we closely examine the cycle diagram proposed by Gartner, it seems that we are nearing the final stages of the peak of inflated expectations and are slowly moving into the trough of disillusionment.
This interpretation of the current market indicates that in the short term, there will be a period where we start to become disillusioned with what's happening in AI. The real applications or use cases for AI will come only after the hype has subsided and the initial excitement has worn off. However, this time might be somewhat different. Fundamentally, AI has the capability to mimic cognitive work, a feature that no previous technology has managed to achieve without human intervention.
That’s it for now. If you’re interested in reading more about the stages of technological cycles, here’s a link to an article I recently wrote.
AI may have a few minor uses. I'm concerned about its potential destruction of creativity though. There is a general sterility to ai products that is difficult to describe. Also, I've found it to be even more profoundly inaccurate on deeply important life questions than Wikipedia. Thus, I call it Wikipedia 2.0.