HaoTech
KEY POINTS TO REMEMBER
Is artificial intelligence (AI) truly a game-changing technology, comparable to the discovery of fire? Or are we blindly plunging into an AI bubble, reminiscent of the infamous DotCom boom and bust of the early 2000s? The issues of hallucinations and the quality of training data are just a couple of the challenges that need to be addressed before we can confidently proclaim AI as a world-altering innovation.
The level of enthusiasm surrounding artificial intelligence (AI) has reached an unprecedented high, even by the tech industry’s customary hyperbolic standards.
Google’s CEO, Sundar Pichai, has awarded AI the ultimate compliment, stating that AI technology is ‘more profound’ than the invention of electricity or the discovery of fire.
Simultaneously, the renowned research institution, McKinsey Global Institute, predicts that AI could potentially boost corporate earnings worldwide by up to $4 trillion annually in the upcoming years. Major players like Microsoft and Google are in a competitive race to incorporate generative AI into the applications and operating systems we use daily.
But what if AI isn’t the ground-breaking, profit-enhancing powerhouse that everyone anticipates?
A number of prominent cynics argue that AI may not yield the expected quick returns – suggesting that the current frenzy around AI might be a bubble.
Is there any merit to their skepticism?
Is AI Overrated? Could It Be a Bubble About to Burst?
CRUCIAL POINTS TO REMEMBER
Michael Hartnett, an investment strategist from Bank of America, suggested earlier this year that the ongoing buzz around AI could be a ‘baby bubble’.
AI-related stocks such as Nvidia, a company whose chips are used in AI applications, have been driving the surge in the Nasdaq 100 this year. Nvidia’s earnings have reached $13.51 billion, a 101% increase from the previous year.
Hartnett has likened the enthusiasm surrounding AI-related stocks to the dot-com crash in 2000. This event saw investments in the Nasdaq Composite increase by 800% before plummeting 740% by 2002.
Economist David Rosenberg concurs, characterizing the current fascination with AI as ‘mania of sorts’. He expressed in a column, “This type of corporate behavior is not too different from what took place in the dotcom bubble.”
However, Rosenberg clarified that he believes in the long-term benefits of AI.
The Issue of AI Hallucinations – Why Are They Significant?
Certain skeptics have expressed doubts about the technology itself. AI analyst Gary Marcus stated that hallucinations, where AI systems like ChatGPT generate or distort facts, are not a minor issue to resolve. Marcus thinks this persistent problem could prevent AI from yielding the financial returns its proponents predict.
Marcus informed the Financial Times, “There is a fantasy that if you add more data, it will work. But you cannot succeed in crushing the problem with data.”
Marcus cited earlier overly hyped AI technologies, such as Facebook’s AI assistant M or IBM’s Watson, which promised revolutionary results but fell short.
Marcus wrote on his Substack, “Hallucinations are in their silicon blood, a byproduct of the way they compress their inputs, losing track of factual relations in the process.”
Marcus believes that the issue of AI hallucinations will eventually be resolved, but it’s uncertain whether the technological breakthroughs needed to solve the problem will come in months, years, or decades.
Can AI Enhance Productivity?
The promise that AI will increase worker productivity has been a major selling point in this year’s AI gold rush. But is it certain that technologies like large language models (LLMs) will enhance productivity?
While McKinsey has predicted that up to 50% of tasks could be automated by 2030, skeptics like technology consultant Jeffrey Funk argue these gains will likely be slower than anticipated.
Funk believes that currently, AI tends to concentrate on tasks, rather than ‘systems thinking’.
The Pitfall of Recursion
The internet is already teeming with AI-produced content, and some warn that this will make future AI training more challenging.
A paper published this summer, ‘The Curse of Recursion: Training on Generated Data Makes Models Forget’ (PDF), demonstrated that large language models trained on AI-created data deteriorate over time. Consequently, training AI models on the entirety of the internet will become increasingly unfruitful.
Ross Anderson, a professor of security engineering at Cambridge University and the University of Edinburgh, remarked, ‘Just as we’ve strewn the oceans with plastic trash and filled the atmosphere with carbon dioxide, so we’re about to fill the internet with blah.’
The Final Word
Is the invention of AI, in all its variants, as transformative as the discovery of fire, or at least capable of altering the course of humanity? Only time will tell. But it’s certain that changes to our lifestyle and work
are occurring at a rapid pace.
Is the AI boom exaggerated? Investors have recently been extremely enthusiastic, and there are known limitations to the technology. Therefore, it might be prudent to maintain a level-headed approach towards AI. Not every AI startup is destined to become a billion-dollar behemoth.
Despite its shortcomings, with investments flowing into AI and business leaders eagerly embracing its potential, AI seems primed to revolutionize our lives in a way not witnessed since the birth of the internet era.
