Finding a balance between innovation and responsibility is vital as artificial intelligence becomes more prevalent.
Key takeaways:
- Artificial intelligence is gaining popularity in the insurance industry
- This technology makes it easier to provide quotes
- Biases can get into the data and lead to discrimination
- Transparency is the key to maintaining customer trust
As you’ve probably heard, AI is taking the world by storm. Tools like OpenAI are available and free for anyone to use, so artificial intelligence is getting more attention now than ever before.
However, many businesses have used this technology for years to analyze data and assist with the decision-making process. For instance, numerous companies in the insurance industry leverage AI to help them develop rates, a practice that isn’t without controversy.
The issue is that there could be algorithmic bias when using AI to set policy pricing and coverage, which isn’t fair to some customers. Here’s a look at what insurance firms can do to address ethical concerns and ensure fairness moving forward.
The ethical challenges of AI insurance
AI ethics is a tricky topic because the goal is to avoid unfair discrimination as much as possible. However, AI can’t reason on its own, as it relies on data inputs, procedures, and processes to create a predictive customer risk profile.
AI algorithms can develop biased predictions that offer better rates or coverage to people in certain groups while discriminating against others. Those predictions often rely on preconceived notions instead of facts, making them biased.
Addressing these issues honestly and transparently is crucial for maintaining customer trust. The industry could also see increased regulation if the use of AI becomes heavily discriminatory.
The bias in AI decision-making
There’s a misconception that AI just looks at data and produces an unbiased result. That’s not necessarily the case, though. For one thing, there could be inherent bias in the data. And for another, AI bases much of its analysis on how similar decisions have been made in the past. So if there was bias in those decisions, it will show up in the AI results, as well.
For example, insurance customers fill out questionnaires containing variables that are used to develop their rates. And some of that information could be used to discriminate. The customer’s zip code, for instance, could impact how much they pay for insurance. While this information seems relatively straightforward on the surface, a zip code can be used as a proxy for race, creating the potential for unfair discrimination.
Of course, AI doesn’t know the customer’s race, but it could still produce a biased profile that quotes a higher rate than someone who lived a few blocks away would receive.
Addressing algorithmic bias
The first step in addressing algorithmic bias is recognizing that it exists. Setting up an algorithm and letting it make decisions without checks and balances is a surefire way to end up with unfair discrimination. Insurance agencies must develop ethical guidelines and try to weed inherent biases out of the technology as much as possible. It’s also important to continue checking the algorithms to address issues before customers begin experiencing problems.
Overall, companies should be using AI to make the process more straightforward for customers. Letting artificial intelligence make decisions without any checks and balances is a recipe for disaster. Once bias creeps into the data, all sorts of problems will emerge. And for that reason, continually auditing and refining the algorithm is a crucial aspect of the equitable use of AI.
Ensuring fairness in pricing and coverage
It’s worth noting that any AI technology the insurance industry uses will discriminate on some level. After all, if there wasn’t any discrimination, everyone’s policy would be identical. The issue arises when this discrimination is unfair and based on things like race or socioeconomic status.
For instance, firms offering life insurance want their AI algorithms to discriminate based on age and health status. Older individuals or those with underlying conditions have to pay more for coverage because they fall into a higher-risk category. That type of discrimination is fair because it influences the insurer’s risk exposure.
The problem arises when the discrimination isn’t based on facts that could change the insurer’s risk. If an insurance company increases a customer’s premiums based on race, that’s a form of unfair discrimination. An AI algorithm without the necessary checks and balances could use a zip code as a surrogate for race and alter the customer’s premiums as a result.
Ensuring fairness starts with looking at biases in the algorithm and addressing problems as they arise. Insurance companies should also have human employees looking at the AI models to ensure they’re comfortable with them from an ethical standpoint.
And this isn’t just an issue of customer satisfaction. In 2019, the New York State Department of Financial Services investigated the use of unconventional data sources by AI algorithms and predictive models in the insurance industry. The Department ruled that insurers should not use external data sources when developing rates and coverages unless they can guarantee the algorithm will follow New York Insurance Law and won’t discriminate against protected classes of people. Insurers must also disclose their data sources to customers when there’s an unfavorable underwriting judgment.
AI and the future of insurance
Addressing the ethical issues surrounding the use of AI in the insurance agency will take some time. The technology remains relatively new, so unfair discrimination is bound to appear as insurance companies work out the kinks. The most important buzzword as companies navigate this transition is transparency. Customers need to understand how their data is being used – and what steps are being taken to protect it.
NICRIS Insurance offers life, home, auto, and commercial insurance in New York. We’re committed to maintaining transparency for our clients, ensuring you receive a fair rate for all your insurance needs. Contact NICRIS for more information or to get a quote.