As AI is increasingly used by companies to improve their products or services, it is up to the leaders to manage and mitigate its associated risks. Although there are various risks associated with AI, some of the most critical ones are the reputational, regulatory, and legal risks companies are exposed to because of ethical decisions made by AI. AI systems must routinely make decisions that are not integrated into their models resulting in a moral predicament.
For instance, a business adopts an AI-based system to optimize the advertisements visitors see on its website. The AI system may include and even push controversial content that forces visitors to express their opinions by commenting. In this case, even though the AI system has achieved its aim of getting visitors to spend more time on the business’s website, there was little or no ethical consideration. It will inevitably drive some visitors away because of the type of content being posted.
Examples of Artificial Intelligence Ethics Issues
One of the most common AI-related ethical issues is in case of a business choosing to automate its processes and eliminating the need for (one or more employees). It raises the question as to whether the company has an obligation to the employee or society? What is the impact of AI on employment levels? Is someone measuring this impact and ensuring that the benefits outweigh the problems?
Now consider that you have applied for a loan to a bank which uses AI to determine your eligibility. If the loan is denied and you are not offered a justification for why it was denied, you will not be able to determine if the decision was ethical or not. (see How AI Can Go Terribly Wrong: 5 Biases That Create Failure)
If the data used for training an AI system did not incorporate adequate information about specific individuals, it will not learn how to handle them. Does an automatic check-in system at a hotel (using facial recognition) recognize a person with freckles? What about a person with vitiligo? What needs to be done in such cases? How can such problems be addressed in a fair manner? (see Why Are Technology Companies Quitting Facial Recognition?)
In case the developer(s) responsible for vetting the data to be used for training an AI system is not aware of potential bias, how can s/he prevent an ethical quandary? If a business has a history of hiring a higher number of men than women, its resume databank will most likely have an inherent bias especially since men and women use different language in their resumes. If the data is derived from men’s resumes, the system will view women’s resumes less favorably.
Ethical Principles for Companies
Companies like Google, Facebook and Microsoft are realizing the importance of such ethical issues. Although these initiatives have had their own problems, lets review some of the positives of what they are doing for AI ethics. Although each company has their own set of challenges, following are some of the lesson’s businesses can learn from them:
Fairness: AI systems should aim to treat all individuals fairly and avoid all types of bias.
Inclusiveness: AI systems should aim to inculcate everyone’s concerns and not discriminate them because of cultural differences.
Reliability and Safety: AI systems should offer reliable results and avoid unintended consequences for both businesses and individuals.
Transparency: AI systems should be transparent and easy to understand and train on.
Privacy and Security: AI systems should be secure and respect the user’s privacy.
Accountability: AI systems should account for appropriate human interaction and control.
Although there is a long way to go, these tech titans are paving the way for addressing ethical AI challenges. What are your businesses doing to address these questions?
Following are four AI ethics recommendations you can integrate in your business now:
- Assign the responsibility of AI ethics to your board.
- Evaluate the various AI-ethics’ risks and devise a plan to address them.
- Involve senior management in the whole process from consideration to implementation to ensure that they are cognizant of how the AI system impacts users.
- Ensure that an element of ethical consideration is built into the system rather than treating it as an afterthought.
Anticipating and addressing these issues will significantly reduce the risk of unintended consequences of decisions made by AI systems. (see AI Can Be Dangerous – How to Reduce Risk When Using AI) Are you mindful of the reputational, regulatory, and legal risks associated with the ethics of your AI in your area of business?
This is a summary of the original Forbes article. To read it, click here.