The Ethics of AI: Bias, Privacy, and Responsibility (What Everyday People Should Know)

The Ethics of AI: Bias, Privacy and Other Issues

Meta description: From the decisions it is tasked with making, to the people it has an impact on, AI can affect all of our lives. Here’s an easily understandable guide to AI ethics, and what you can do about it.

Imagine applying for a job that you’re perfect for, and subsequently never hearing anything back from them. Imagine also learning later on that an automated system had dismissed your application before any humans saw it. Nobody set out to treat you unfairly, they just happened to be using a system that wasn’t prepared for people like you.

These are the types of situations that the concept of AI ethics addresses: not evil robot overlords, but everyday decision-making by automated systems and algorithms. You don’t need to know any technical background on AI to care about ethics in this space; here’s some background information to understand what problems there are and what the solutions might be.

Bias In AI

As this article series has discussed in the past, AI systems are generally trained on information gathered by humans. If that information reflects human biases, then the AI will also display them. Since there are many different areas of application, there are many different potential issues here. For example, as has been reported in the news, some facial recognition technology has had substantially greater difficulty with certain ethnicities. Some hiring or loan approving algorithms may be more likely to reject certain populations. The specifics change for each situation, but the problem remains the same: the use of information that humans can be prejudiced with is prone to make decisions with prejudice.

Some solutions include

training AI systems on diverse and representative data sets,

making sure systems are tested on different demographics before being deployed,

and

having humans involved in decision making and in disputing automated decisions.

Privacy

Every time you use an app, search something up, or utilize an AI in any way, there is potential for it to collect information about you. Some of that information can be potentially sensitive things like your location, search habits, health, or physical appearance.

Some questions you might want to ask include:

What information is being collected about me?

When does this information get deleted, and who can see it?

Is my personal information going to be used to train other AIs?

Can I deny or delete this information?

Some ways you can protect your own information include

not giving private information to AIs that don’t need it, such as passwords, social security numbers or other private information,

looking into privacy settings, as many AIs will allow you to delete your information or opt-out of being tracked,

reading the fine print and researching the company, and looking for a general summary of what information they collect and for what purpose, and

being wary of AIs that request too much information, particularly biometrics such as facial recognition or voice scanning.

Laws regarding privacy and what companies are allowed to do with your personal information can vary wildly depending on the country you’re in, so you may want to look into what specifically applies to you.

Accountability

In the event that an AI does something wrong, like give bad advice or make a prejudiced decision, who is responsible? This can be a complicated question with no easy answers. The laws in many countries are currently in flux on this issue, and different governments and organizations have different stances. What most people seem to agree on is that there needs to be some level of human accountability for the actions of AI. AIs should not be able to act entirely independently when it comes to important decisions, and if they do, humans should be responsible for reviewing and overseeing their actions.

Other Ethical Issues

Misinformation and deepfakes are a growing problem for all mediums, and AI makes it easier than ever to generate fake photos, videos, and audio. As a result, people in general should be extra wary of surprising or unbelievable content, and fact check before sharing.

Copyright and creator issues are also significant, in that many AIs were trained on content generated by humans who didn’t give permission for it to be used. This can apply to text, art, videos, and more. There is an ongoing discussion about whether or not content creators should be compensated for their work and time in this area, but laws are still being ironed out.

Jobs and ethics intersect in a few different places: questions about retraining programs, about who profits from AI, and about whether its use is fair.

The environmental impact of AI is also an ethical issue due to the amount of energy it takes to run and train.

What Can Regular People Do?

You probably can’t believe everything you read or experience, and as mentioned above, there are ways to fact check.

It helps to be a skeptic: if something is presented to you by an AI, ask yourself where it could have come from, and whether it is reliable.

You should also verify things before sharing them, or using them in an important way, and take care to use AIs with known good privacy policies.

In terms of disputing AI decisions, where possible, go through official channels to appeal a decision you think is incorrect. Many systems now require a human to review a disputed decision. You also have the right to not use a specific company’s services if you don’t like their practices.

As far as using AIs responsibly, it helps not to misuse them to spread misinformation, engage in harassment, scams, or other malicious activities. AIs can also be used to impersonate you or record your voice or face without consent, which again is a form of harassment.

What Can Business Owners And Creators Do?

If you are using an AI for your business or creative work, it’s a good idea to be up front with your customers about what AI processes were used. If applicable, make sure to have a human review work before it is delivered or published. Take care to not collect too much information about your customers, and delete it if you do. Review things that your AI produces to make sure they are factually correct and not prejudiced.

A Balanced View

None of this is to say that AI is inherently bad- there are many applications where it is useful, helpful, and saves time and money. Medicine, language translation, accessibility features, and more all utilize AI to help people. That being said, ethics in AI exist to make sure that AI is being used responsibly, and that decisions that impact real people are made with care and without prejudice. The potential pitfalls shouldn’t overshadow the practical uses, but precautions should be taken wherever possible.

Like with cars, laws and ethics around AI are in the process of being worked out. Seatbelt laws didn’t come about because cars are dangerous, they came about because they are seen as necessary to keep drivers safe. Similarly, many laws around AI are focused on making sure that people are protected against its misuse.

Conclusion

All of that being said, the single largest ethical question around AI is what should be done with it and how to make sure that nobody abuses it. This is a conversation that everybody can take part in, and you don’t need any technical expertise on AI to contribute. I hope this series has provided you with some of the background information you need to be informed, and that you will engage in these discussions. If you are looking for something more specific, you can look into the provided links and resources. What are your thoughts on the ethical questions surrounding AI? Which of these issues do you think is the most pressing? Feel free to comment below!