How to overcome ethical challenges in AI? Commentary to The Telegraph and advertorial piece MsfT

Lawyers are quick to describe differences between ethics and morals in that you can define laws for ethical behavior and a “code of ethics” but not morality which is a matter of subjective judgment and personal values. This is a complex issue that becomes much harder in the case of machine learning algorithms that define coding policies of action from automation. Lawyers again point to defining “jurisprudence” issues of who is legally liable as a human and not a machine over a machine action in accordance with human wishes or ethics. Creating a clear undisputable line between ethical use and misuse remains a hard problem in AI because while simple fully controllable and observable events such as “stop or go” or calculate an "optimal answer”, this becomes vague when in real-world problems of social news free speech to deciding who gets health insurance to surveillance and weaponization in how to decide the rule to say yes or no this is acceptable.

What we have seen companies in the corporate world doing is to create a charter of AI conduct such as Microsoft’s Code of AI to define company guidelines for personal and customer data privacy, fairness, inclusion, transparency and safety, and accountability. What is at stake here is “Trust” and being “trustworthy” n protecting reputation and behavior. This is the first step but it still needs ethical use of AI in operations in practice and companies to need to also introduce advanced “data management practices” to measure and define the visible use of data to guard against the machine-learned bias that directly or indirectly can skew what may be well-intentioned automation and customer services.

A more concerning issue is the weaponization of cybersecurity hacking that creates deliberate ways to “poison” machine learning training data to the rise of deepfake news and images. This means most companies will not have the skills to combat these new advanced threats described in my recent book Navigating new cyber risks We are in an arms race to defend against these threats and ethical accountability to hold hacker criminals and rogue governments to account but equally to insist on due diligent of ethical conduct by corporations to manage customer data and its secure encryption today and the future. With the rise of quantum computing new forms of quantum key security are in need to protect today's privacy and data as in the next decade this data will be hackable by future adversaries. A whole new mindset is needed to think about ethics and how digitization transforms lives and liabilities.

  • Black Facebook Icon
  • Black LinkedIn Icon
  • Black Twitter Icon
  • Black YouTube Icon

© 2020 Artificial Intelligence Innovation Network, Warwick Business School, UK