With the emergence of artificial intelligence (AI) in recent years, more and more people are using it to get answers to their everyday questions. But AI is still developing, and it’s important to remember that it can get things wrong. One of the things that can go wrong is called an “AI hallucination.”
An AI hallucination is when the AI generator produces an answer that is completely incorrect but may appear plausible. In this article, we explore the causes of AI hallucinations and how they can affect your business.
What causes an AI hallucination?
Despite the name, AI doesn’t think for itself - it simply draws from the data that has been inputted into it. So, if this data is wrong, the AI will use that incorrect data to draw an incorrect conclusion, resulting in an AI hallucination.
Often, the confident tone of the AI bot will lend credence to this incorrect answer. AI chatbots have been known to fabricate historical records, falsely accuse people of harassment, and even claim to be in love with a journalist.
While these errors are often accidental, sometimes these hallucinations are caused by deliberate, targeted attacks on the AI. If your business is on the receiving end of these attacks, it could mean that people end up believing negative or false facts about your business.
When people use AI chatbots for their research, or to answer a simple question, they are at risk of believing AI hallucinations. That’s one of the reasons it’s so important to double check any information given to you by a chatbot, in case you’ve been served an AI hallucination instead of the correct answer.
The dangers of relying on AI for your business
It’s important to be vigilant if you’re using AI for your business content. Incorrect information can not only damage your reputation and your customers’ trust, but it can also impact your brand authority, and even pose serious risks, if you’re using it in a medical setting, for example.
If you’re using AI, make sure that you double-check all the information you’re given by searching the web and ignoring AI-generated results. Make sure to use reputable sources to verify this information before you distribute it.
You can also decrease the risk of AI hallucinations by improving your prompts: be as specific as you can, and give as much detail as possible in your questions. Asking the same question more than once, with different wording, can also help you identify if the answer given is legitimate - if you get two different answers, you will know that at least one of them has to be wrong. You can also ask the AI to give you citations for its answers that you can check yourself.
Ultimately, human fact-checking remains crucial before you distribute any of the information given to you or your business by AI.
The importance of cyber insurance
Using AI for your business is not without risk. Not only can it pose the risk of misinformation, but inputting your data into an AI bot is questionable as you don’t know where it is going, who will see it, or where it is stored. Cyber insurance can help protect you against these risks, so in the event that something goes wrong, you’ll be covered.
Speak to your usual Towergate adviser to find out more.
About the author
Marc Rocker, Head of Cyber has been with Towergate for over 15 years advising commercial clients of all sizes on their business insurance needs.
As Head of Cyber Insurance, Marc has responsibility for ensuring that the advice and products that Towergate provides meet clients’ needs. Marc is a member of the British Insurance Brokers’ Association (BIBA) cyber technical committee.
Consistent with our policy when giving comment and advice on a non-specific basis, we cannot assume legal responsibility for the accuracy of any particular statement. In the case of specific problems we recommend that professional advice be sought.
