AI Reality Check: 5 Wildest Exaggerations So Far

From SciFi-tinged worries about an AI takeover to claims that LLMs could hack websites without human help, the GenAI hype machine is in full swing. Each morning, a new batch of overheated and unsubstantiated claims about AI capabilities arrives, fouling in-boxes and poisoning news feeds.

GenAI’s arrival is a transformational moment — everyone gets that. But is it really going to destroy democracy, create mass unemployment, or level entire industries overnight?

There’s a growing body of evidence that Current Thing is just the latest in a string of periodic AI furors that started in the 1970s, each one opening with a rush of excitement before petering out on sullen declarations of an ‘AI Winter.’

Are GenAI’s Dog Days just around the corner? We sanity-checked some of the biggest claims to date. This is what we found.

Table of Contents Table of Contents

1. Key Takeaways
2. ‘God-Like Powers’ or Shameless Puffery?
3. AI Hype vs. Reality
4. Adjusting Expectations
5. The Bottom Line
6. FAQs
7. References

Show Full Guide

Table of Contents

1. Key Takeaways
2. ‘God-Like Powers’ or Shameless Puffery?
3. AI Hype vs. Reality
Show Full Guide
4. Adjusting Expectations
5. The Bottom Line
6. FAQs
7. References

‘God-Like Powers’ or Shameless Puffery?

The interwebs are awash with instant influencers who mass-promote AI’s ‘god-like powers,’ which on closer inspection include miracles like Waifu name generators and apps serving up inspirational quotes from noted thinkers such as Notorious BIG.

Less than 2 weeks since OpenAI started rolling out GPT-4o.

And people have been busy executing god-like tasks.

50 truly mind-boggling examples:

(29th is my favorite)

— Sai Rahul (@sairahul1) May 25, 2024

What compels AI enthusiasts to engage in shameless puffery?

• There’s a Lot of Money to Be Made: GenAI firms are the current generation of tech unicorns, generating eye-watering valuations and setting the stage for massive IPOs. Firms that buy into GenAI services or develop AI applications of their own hope to ride the bandwagon, generating new revenue streams and future-proofing their business models.
• Keep Competition at Bay: Google Brain co-founder Andrew Ng has said some established tech players don’t want to compete with new AI entrants and use fear of AI to lobby for strict regulation.
• Fear Sells: AI leaders also use fear to amplify AI’s potential impact, e.g., ‘this technology is so powerful it could wipe out humanity’ (ergo: don’t let your competitors get to it first).
• AI Leaders Can Be Fabulists. OpenAI’s Sam Altman is well known for making breathless pronouncements about AI’s fearsome power that he later walks back. Others overstate the capabilities of their inventions or weave fantastical tales about themselves.

Long before GenAI, Marc Andreessen said that software was eating the world. It was a controversial take at the time, but years of real-world evidence at least gave it a ring of truth.

Today, there’s a whole lot of snake oil being slung in the name of potential AI transformation. Here are some of the biggest overpromises, exaggerations, and sensational claims.

AI Hype vs. Reality

Exaggeration 1: AI Will Cause Mass Unemployment

The Claim: AI-enhanced automation will eliminate scores of administrative, finance, operational and management roles around the world, in numbers roughly equivalent to the entire population of the USA.

Any sector or profession reliant on brainwork like law, software development, architecture, engineering, healthcare, or creative industries, will be upended as AI takes on the heavy lifting of any complex but repeatable tasks.

The Context: Earlier this year, Elon Musk took to the airwaves with dire predictions of an economy where humans would no longer be needed. In March of 2023, Goldman Sachs published an alarming report predicting generative AI could replace or degrade 300 million jobs. The figures seemed to give credence to previous predictions about an AI jobs apocalypse.

The Reality: The Goldman study also estimated GenAI’s impact on productivity growth to be circa 1.5% annually over ten years. That would hardly end the need for human workers.

A separate finding suggested that 2/3 of jobs in the US faced ‘some degree’ of AI automation. The truth is, we’ve been here before. Instead of eliminating people working in offices, it’s just as likely that AI will become a productivity tool.

Like automation, smartphones, and collaboration platforms — AI might reshape jobs rather than shred them.

Lots of takes that AI will lead to massive job loss. These takes are identical to doomsday views of industrial automation from past decades.

Technology doesn’t destroy jobs, it transforms them. Problem is lack of investment in supporting and retraining displaced workers.

— Nahaz (@NahazDota) February 4, 2023

Exaggeration 2: AI Will Turn on Its Human Masters

The claim: Once AI achieves sentience, it will quickly take note of humanity’s physical and intellectual inferiority. The obvious next move is to take control, even enslaving or eliminating humans if they are seen as a threat to AI’s existence.

The Context: This Terminator-inspired nightmare exploded again last June when it was widely reported that a trial by the US Air Force had gone tragically awry when an AI-controlled drone had attacked and killed its human operator.

There have already been signs that AIs are capable of hatching their own secret agendas. Meta’s CICERO AI model has been shown to engage in premeditated deception when used to play the Diplomacy board game. Others have worked out how to bluff in poker, conduct a feint in StarCraft II, or deliberately mislead other players in economic negotiation simulations.

The Reality: The drone killing turned out to be a thought experiment misreported as an actual event, a hypothetical delivered in a defense industry presentation.

Examples of AI ‘lying’ are mostly just outputs from an inherent flaw in machine learning algorithms. Sometimes, an AI lacks the information to make an accurate prediction due to conflicting information or poor-quality data. Inferences may be flawed, and the results can look outlandish, but GenAIs aren’t sentient. They can’t ‘lie’ like humans.

An AI-enabled drone killed its human operator in a simulated test "because that person was keeping it from accomplishing its objective." https://t.co/AWKu9KRh4x

— VICE News (@VICENews) June 1, 2023

Exaggeration 3: AI Is Driving Incredible Business Growth

The Claim: AI-powered applications will make employees more productive, make operations more efficient, improve service, identify new business opportunities, and add flexibility and resiliency in times of disruption. Loyalty will strengthen and customer churn will disappear. Thinking digital twins will make offices, factories and transportation systems more sustainable and energy efficient.

The Context: Companies in almost every sector are hoping that AI will transform their business models, lead to amazing new products and services for customers, and bring untold new profits to investors.

A study by IDC and Bank of America predicts that AI could contribute $15.7 trillion to the global economy by 2030. Research from McKinsey suggests GenAI could add the equivalent of $2.6 trillion to $4.4 trillion annually.

The Reality: A few will profit handsomely from the GenAI surge, but who — and by how much? Google and Amazon have reportedly been damping down expectations about AI’s impact on revenues.

In a recent earnings call, Salesforce said its many AI projects won’t make ‘a material contribution’ in the 2025 financial year. As for innovation, a lot of R&D is focused on building better chatbots or optimizing current use cases for automation.

Great news, but is it the stuff of a dramatically re-imagined business future?

After a year of hype, the reality is emerging. Cloud clients aren’t buying generative AI tools because they’re expensive, lack accuracy, and it’s not clear what value they provide. Some analysts are already warning of a coming “trough of disillusionment.”https://t.co/RRxImtQG9j pic.twitter.com/0Px1exFT1t

— Paris Marx (@parismarx) March 12, 2024

Exaggeration 4: AI Сan Read Your Mind

The Claim: Brain implants backed by AI can decode brain waves and discern patterns that can make your thoughts visible. Is the day coming when neural implants paired with GenAI can know your thoughts? Will police detectives soon be interrogating suspects while observing what they think on a lie detector console?

The Context: Earlier this year, Elon Musk’s Neuralink implanted a chip inside the brain of a 29-year-old man who was paralyzed from the shoulders down. The chip has enabled the patient to move a cursor on a screen simply by visualizing each motion.

In May of 2023, US researchers showed they could see the words people were thinking using brain scans read by generative AI. Another project with similar aims generated headlines about the ‘Ai hat that can read your mind.’

The Reality: AI’s can be trained to identify specific brain wave patterns with some degree of accuracy, but the ‘flickering, lightning-fast, multiple-stream’ nature of human thinking means machine-reading our innermost thoughts is a long way off.

And while Neuralink’s invention of a working wet-dry interface is reason enough for acclaim, its thought-to-motion trick doesn’t break entirely new ground. Researchers in brain-machine interfaces (BMIs) have been interpreting human brain signals for at least a decade and using them to control things like robotic hands, arms, computer cursors, and wheelchairs.

AI makes non-invasive mind-reading possible by turning thoughts into text https://t.co/LtuTT40fAX

— The Guardian (@guardian) May 1, 2023

Exaggeration 5: AI Is Becoming More Human

The Claim: In January, OpenAI CEO Sam Altman told an audience at Davos that Artificial General Intelligence (AGI), a future iteration of AI that can complete tasks as well as or better than people, could be created in the ‘reasonably close-ish future.’ He’s also suggested that children will soon have ‘more AI friends than human friends.’

The Context: Enthusiasts regularly predict that the day is coming when AI will think and behave like flesh and blood humans.

Eliza Kosoy, a researcher at MIT’s Center for Brains, Minds, and Machines, said recently that artificial intelligence is ‘already exceeding humans in some domains.’ AI-driven personal assistants are being trained to speak with human intonation and nuanced emotion.

One chatbot using cognitive behavioral therapy (CBT) techniques really does help people reduce anxiety and depression.

The Reality: Generative AI tools are indeed being trained to mimic human behavior more closely, but mimicry is what it is — just a very advanced form.

Chiara Succi, Associate Professor of Organizational Behavior at ESCP Business School, says humans have a unique type of intelligence that AI can’t match:

“AI doesn’t have the sensitivity to manage unexpected events or crises. A self-driving car cannot enter a complex roundabout as it cannot identify a zero-risk condition. Human drivers can adapt to the flow and moderate their speed to avoid accidents, something a robot will probably never be able to do.”

I heard some professor put googly eyes on a pencil and waved it at his class saying “HI! I’m Tim the pencil! I love helping children with their homework but my favorite is drawing pictures!”
Then, without warning, he snapped the pencil in half.
1/2

— Greg Stolze (@gregstolze.bsky.social) May 15, 2024 at 7:31

When half his college students gasped, he said “THAT’S where all this AI hype comes from. We’re not good at programming consciousness. But we’re GREAT at imagining non-concious things are people.”
2/2

— Greg Stolze (@gregstolze.bsky.social) May 15, 2024 at 7:33

Adjusting Expectations

Gartner’s Hype Cycles have become part of tech’s narrative canon, providing a remarkably reliable way to map a popular new innovation’s trajectory from idea to reality.

Last year, AI reached the model’s Peak of Inflated Expectations. Today, it’s on the precipice of the dreaded ‘Trough of Disillusionment,’ where overheated promises give way to cold, hard facts.

Rather than get depressed, AI boosters should welcome a period of calm and clarity. Because too much puffery comes at a cost.

People feel tricked, money is mis-allocated, enthusiasm cools and sarcasm sets in. Investors go looking for safer bets — or the next big thing.

The Bottom Line

The #rocketai hoax of 2016 still provides the best cautionary tale of how tech enthusiasm can dull the critical faculties of academics and hard-nosed business people alike.

Time again to balance natural positivity for AI’s potential with sensible assessments of business use cases and expected ROI.

FAQs