Skip main navigation

New offer! Get 30% off your first 2 months of Unlimited Monthly. Start your subscription for just £35.99 £24.99. New subscribers only T&Cs apply

Find out more

The ethics of AI – Failure

AI can still experience failures and lead to unintended consequences, watch as Marcus Liwicki explains in this video.
Hej, and welcome! In this session, we want to speak about ethical aspects of AI.
Before I start with defining what ethical aspects we actually want to consider when we create AI, I want to give some motivation why it is important to speak about ethical aspects. Maybe some of you remember that, for example, a few years ago, seventy thousand people here in Sweden did not get their unemployment allowance due to a fault in an algorithm of AI. And also, there was a rather dramatic failure IBM Watson Oncology, a system which can maybe cure cancer or at least predict the measures to be taken in order to cure cancer. It did some wrong decisions and gave dangerous treatment, advises. The algorithm was trained on two small data and typically
data were: quite severe current cases of cancer were there, and it was good that it was only a human assistance system. So the doctor finally made the decision and it only gave some advise. Good aspects, of course of that systems are. Could maybe reason why it came to some conclusion and then the doctors could object. Another quite famous example is the Chatbot “TAY” that’s actually already five years ago, but maybe you heard about that. It’s a chat bot which was developed by Microsoft and released, - one would say “too early” - because it was like picturing, I think, 16-17 year old teenage girl speaking about how nice the world is.
So just text messages I like lying in the sun and whatsoever. But it was trained to interact with the humans. So if humans say something, it would maybe take over some ideas and some of the concepts of humans say Humans tried out some not so nice things like I love Hitler and whatsoever. And after 24 hours, they had to switch off the chat bot because it was so corrupted, being quite freaky Hitler loving girl. And that was quite dangerous.
And one saw, yeah, if there are really no ethical concepts integrated in the AI and there are people around who want to make it harmful then the AI can become quite dangerous tool and actually in other areas, also quite dangerous weapon. Another aspect was an AI for recruitment used by Amazon, and they used all the applications from the past and identified which could be the top candidates. However, most of the workers at Amazon at that time in the tech team were white young males and then application from other people who were not male and not white, and maybe not young.
They didn’t have any chance, even though they were highly skilled, because AI was biased to predict good person just based on the simple features white, young and male. So there was a very big, unwanted bias in the AI, and all of these aspects we have to consider.

Interesting real-life AI failures scenarios in ethics from big IT giants were shared in the above video. AI failures lead to critical issues in different aspects. These examples will help you to understand the importance of considering ethics in AI.


  • The sensitive relation between ethics and AI implementations.
This article is from the free online

An Introduction to Artificial Intelligence in the Tourism Industry

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now