I am not an expert on what’s going on in the field of artificial intelligence. As a professional futurist, though, I am trained to see and think about what others may not be seeing. To this end, I have four observations that may prove helpful to AI investors.

First, all the leading AI players are now investing heavily in the underlying technology, personnel and data centers because they are convinced that whoever reaches artificial general intelligence (AGI) first will have an incredible competitive advantage over everyone else.

I understand the logic, but doesn’t it then also stand to reason that those who are not first will be at a competitive disadvantage? If true, the fortunes of those coming in second place could plummet. 

Second, AI is now capable of producing its own “synthetic output.” In other words, AI is producing original insights. Much of this synthetic output may yield extraordinary new insights and could even lead to the creation of beneficial developments in the fields of drug discovery, logistics, and material science.

The risk, however, is that a small fraction of the synthetic output will be inaccurate. If this faulty output is not immediately caught and quarantined, next-generation AI models may “vacuum up” this tainted output and use it to produce insights and findings that are even more tainted. This negative feedback loop could quickly cascade out of control.

As an analogy, consider the famous Tacoma Narrows bridge collapse. The faulty construction of the bridge, in combination with high winds, created a vicious feedback loop that brought down the whole bridge. I fear something similar could happen with today’s AI models.

A third risk is that AI will produce extraordinary gains in productivity, but these gains will come at the expense of workers. If the economic downside of high unemployment is greater than the upside gains in productivity, the result could be a recession and the entire economy could soften.

A final risk is that society may not embrace AI as willingly as its proponents might lead us to believe. The massive amount of “AI slop” being produced is already making it more difficult to navigate the Internet, and, soon, AI-produced “erotica” will further flood the digital world. Is it not possible that the millions of people weaned on the promise of smartphones and the liberating freedom of social media–but who now better understand the damaging side-effects of these tools–may be more cautious before adopting AI?

I am not a financial advisor but I encourage investors to diversify their investments in AI, assess its risks more fully, and understand that much of the world may view the promise of the technology differently than you. Forwarded is forearmed. 

Jack Uldrich is an author and a futurist. His latest book is “A Smarter Farm: How Artificial Intelligence is Revolutioning Agriculture.”