How to make AI systems people can trust – a web lecture

Special Series After some years of investigation, more and more industries are warming to integrating machine-learning technologies into their digital offerings. Given the almost omnipresent reports on biased algorithms and security concerns, however, consumers often have an understandably hard time finding enthusiasm for AI popping up in sensitive areas, such as health and finances.

A good step towards curbing acceptance is to build systems in a transparent manner. This means they should ideally provide both insight into the data used to train a model as well as ways of reconstructing how an algorithm arrived at a certain decision. But this is easier said than done.

The third episode of our MCubed web lectures on practical machine learning will therefore zoom in on the things to keep in mind when implementing explainable AI systems.

On November 4 at 11am UTC (12pm CEST) Napier’s Chief Data Scientist and all around brilliant person Dr Janet Bastiman will join the series to take a look at where legislation is headed, how we got to this point in the first place, and what we can do to make sure that our end users have trust in our systems.

Dr Bastiman will also help you figure out what you can versus what you should do, and tackle topics ranging from principles for algorithmic accountability and data transparency through to making explainability part of the development process. After all, it’s not just end-users profiting from transparent systems, it also helps those on the team who need to keep models in check and makes things easier if compliance is questioned.

If you haven’t run into Dr Bastiman on one of her frequent speaking gigs, you’re in for a treat with this episode.

Besides being an experienced data scientist, Dr Bastiman is a committee member for the Royal Statistical Society Data Science Section and holds degrees in molecular biochemistry and mathematics, with a PhD in computational neuroscience. With over 20 years of experience, she’s more than familiar with the peculiarities of data science in telecommunications, marketing, and the financial sector, and knows what’s important for startups and established businesses when it comes to implementing and improving their AI offering.

So if you want to learn more about how to win trust and course-correct models, join us on November 4. Oh, and the offer still stands: the MCubed webcast series aims to provide machine-learning practitioners with actionable advice for their day-to-day work, so if you have questions on anything from model serving to troubleshooting in production, give us a shout and we’ll make sure to incorporate it in one of the upcoming episodes. ®

Source link

Related Articles

Back to top button