Last July, I had the privilege to attend a panel discussion on Ethics in Artificial Intelligence. It was an interesting discussion and I felt that such discussion can be pretty abstract for the non-technical folks.
There are many cases where AI has been misused intentionally (such as DeepNude) and unintentionally (biased job applicants selection ). At the end of the day, we need to realize that AI is a tool and how it is used, for harm or the greater good really depends on the individual. As Uncle Ben once said,“With great powers (coding, machine learning and artificial intelligence), comes great responsibilities”.
Here are my thoughts after the panel discussion.
Most governments have prepared guidelines on using Artificial Intelligence by now. These guidelines will definitely form good “reference” materials for any companies or individuals who wants to understand more about ethical usage of Artificial Intelligence. Organizations looking to prepare their guidelines, will need to start arming themselves with the technical knowledge of AI, how it works and may work, what are the near-term possibilities and what are the “impossibles”. But guidelines at the end of the day can only reduce the “unintentional”. To reduce the “intentional”, society needs to have a strong regulatory and law framework, requiring any regulators to have a thorough understanding on how AI works otherwise we may be punishing the wrong actors.
Second thing, if one has observed history, during Enron and Tyco saga (early 2000s), where accounting processes were abused, there were discussion on Ethics in business managers. Accounting process was created to help management and investors understand in detail where the business is heading. It was never meant to paint a false picture.
Moving on to 2008, during the Financial Crisis, again there were discussion on Ethics, this time round on the bankers. Complex investment products were pushed and sold to the public, business organizations and pension funds, resulting in huge losse. Retail investors were hit the most and now having to rebuild their retirement nest eggs.
In this third rise of the AI, we are now moving on to say that there needs to be Ethics in ML and AI community (the engineers and scientists). This time round greed is not the cause anymore but rather ignorance, which is not difficult to overcome. We just need more education and critical thinking to avoid negative usage of AI.
I believed there are two things that we need to do right NOW:
- Educate the public on what Artificial Intelligence and Data Science is about, what kind of information can we (Data and AI Scientist) get from your data. As individuals, we need to start realizing that our data has value and it can be abused if not protected well.
If you are in Singapore, a good start is to go for “AI 4 Everyone” conducted by AISingapore. Checkout their “Events” page to find more details.
If you are not in Singapore, start looking out for tech communities, listen to their talks to understand what they do with data, ask questions if needed. If all else fails, Google is your best teacher. :)
- We, the AI and DS community need to start realizing that our work can have an impact on our customers and general public. We have to start asking ourselves if the impact that we are creating, although may have helped the business overall, is there a "harmful" impact to the other stakeholders as well, namely the customers.
There are many places that can go ‘wrong’, for instance biased data, models that are extremely difficult to interpret (like deep learning, random forest, support vector machine), and how decisions are made in high stake situation such as healthcare and education.
The tech community should start putting ourselves in the shoes of the customers, and empathise how the customers feel at the “receiving end” of the models that we build.
At the end of the day, I do not think Ethics is just applicable to the tech community (at the moment) but to everybody; to start thinking more about the impact of our actions, both personal and at work. The global conversation/discussion on ethics in artificial intelligence should definitely start now to raise awareness, both in the public and the tech community and it will definitely take time. Another important point to note is that these Ethics discussion will never have a conclusion. But having a conclusion is not what we want to achieve but rather having the awareness, knowledge and empathy to consider from many angles/views.
I hope the post has been useful to you. Feedback and discussions are welcome as long as we take the approach of NOT CONVINCING but more of considering different viewpoints. :)