New Zealand needs more transparency about how algorithms and AI reach decisions and accountability to ensure they are operating ethically. We also need an AI watchdog, says Jane McCarroll.
How do we protect the robotic workforce from pre-programmed bias?
We live in a world which is inherently biased. Humans have innate bias and there is a body of thought that claims even big data is biased. When you look at what’s going on in AI it seems very likely that the biases that already exist in society will be exacerbated or reinforced. Yikes.
There are issues around gender bias. Who gets the interview, the loan or the policy? This phenomenon could also amplify other biases in data, for example, issues around race.
Machines learn their prejudices in language. If a bias is detected in an AI algorithm, it is easy to rectify this simply by reprogramming it. It is getting people to acknowledge the bias in the first place that is the hard part.
Algorithms were supposed to free us from our unconscious mistakes but now there is a new set of problems to solve.
How do we address the potential for discrimination in an incredibly complex environment that is already quietly embedded in our personal lives and in some of the most powerful institutions on earth?
AI is such a new field where these issues are really not being discussed enough. There are also questions of the ethical implementation of AI.
We need men and women to be shaping our technological future, and we need to have a diversity of thought in how we shape our technologies.
Here are three areas where I think we need to dig deeper to address our challenges.
We need a governing body: Algorithms can make bad decisions which have serious impacts on personal lives. New laws and better government regulation could be a powerful tool in reforming how companies and government agencies use AI to make decisions. We need an AI watchdog. Maybe another Fair Go but for AI.
We need transparency and accountability: We need more transparency about how these predictive tools reach their decisions and accountability to ensure they are operating ethically.
We cannot rely on AI to have the moral innovation to try and improve our lives or our society. As long as the bias exists in humans, it will not only present in AI but the bias could be amplified.
This is not just a technical problem. It goes back to having human discretion and not thinking all the tough questions can be solved by technology. We have already seen disastrous consequences of bias in technology. When film was developed it was originally biased towards white people. The bias towards skin with higher reflectivity meant that there were often exposure issues when shooting mixed-race photos.
It is not pretty. Bias is bias, whether conscious or unconscious.
We (still and always will) need good manners: What does good manners mean to robots? What does ‘accuracy’ and ‘fairness’ mean to AI?
Once we move beyond the technical discussions about how to address algorithmic bias, there is another (digital) elephant in the room. How do we decide what accuracy and fairness mean? If we want AI to be more accurate, what kind of accuracy is most important? If we want it to be fairer, who are we most concerned about treating fairly?
How is this being measured? Or audited? And by whom?
We are entering the 4th industrial revolution. Change is a given. We are already sharing our lunchroom with robots. Let’s do this.
AI needs to exist collaboratively and productively with humans, and vice versa.
We need to encourage more women into the industry. Work already being done in this field needs to continue, and we also need to increase our efforts to reduce bias.
Earlier this year I launched a series of events in partnership with IBM on ‘The Future of Work’. The series was designed to encourage learning, dialogue and development to proactively and positively prepare ourselves for the changes ahead.
We ran three events in both Auckland and Wellington around what the workplace will look like in 2020. What are the skills required? How can leadership evolve and thrive?
Our last event saw a panel of futurists share their vision on how we can lead and deliver AI and automation in our workplaces.
Leading research coupled with shared real life experiences was both galvanising and illuminating. At every event, the questions outweighed the time allocated. There is an unsatisfied hunger to learn more. Watch this space. M
Jane McCarroll is the marketing and membership manager at IMNZ. The Institute of Management NZ helping leaders step up and lead since 1946.
