AI and the ghosts of tech booms past

The Bank of England has warned that a crash in AI could potentially lead the UK into economic recession, estimating a potential 2.2 percent drop in GDP if there's a "price correction" in AI stocks due to shifting productivity and profitability within tech-driven companies. This should serve as a cautionary tale for businesses worldwide.

However, the risks associated with AI extend far beyond Silicon Valley, and many are currently asking the wrong question: "Is AI real or is it hype?" The last three decades of technology have taught us that the answer can be both.

The UK experienced a similar crisis with the Y2K bug in the 20th century. The government spent billions preparing for the millennium bug, testing and patching systems that underpinned various sectors such as banking, benefits payments, air travel, and the National Grid. The extensive efforts to mitigate the risk paid off, as the world continued functioning as usual after the midnight deadline.

This experience demonstrates that what may appear to be an exaggeration in hindsight could have been a genuine risk that was effectively managed.

Cybersecurity also shares a similar risk profile. A well-managed cybersecurity threat can be perceived as overblown in retrospect. The dotcom bubble teaches a different lesson, where valuations often did not reflect the true value of companies with little profit or questionable business models. Despite the bubble bursting, the underlying ideas proved to be valuable, and we now use the internet every day.

Crypto introduced a third lesson, where revolutionary language could mask weak use cases, with speculation, celebrity endorsements, and a misguided pursuit of purpose.

AI's rise mirrors these three stories. It carries risks that may seem exaggerated until they are not, such as cybersecurity and governance issues. Organizations must address these risks by treating cybersecurity and governance as operational disciplines rather than side projects. They should focus on understanding data input, system access, authority, potential manipulation, and response strategies for potential failures.

Businesses should also invest in capabilities, data foundations, people, and workflows rather than betting everything on the most popular AI vendors, models, or products.

The current moment presents a challenge for business leaders, as AI could be both a real general-purpose technology and a bubble. Historical examples show that hype and substance are not mutually exclusive. Real risks can be exaggerated, while transformative technologies can attract irrational investment and powerful ideas. The sensible approach to AI is disciplined experimentation, viewing it as a portfolio of bets, including defensive, exploratory, and skeptical strategies.

Companies must ask critical questions about AI's purpose, users, data access, system influence, authority, and contingency plans for failures. By adopting this balanced approach, businesses can navigate the complexities of AI and avoid repeating past mistakes.