On the invitation of Philippe Warzée, editor-in-chief of Pub.be, I was pleased to present my latest thoughts on the impact of artificial intelligence in the media, retail and insurance sectors.
The presentation started with a brief historical overview of recommendation algorithms, perhaps the most widespread type of algorithms around. After this introduction I dealt with 3 use cases : Netflix (media sector), coupons in the retail sector, and health insurances.
I followed with an analysis of what had changed in the last 25 years in terms of data collection, prediction models and techniques used.
Finally I concluded on the risks posed by current techniques, how public service can change the picture and directions for making algorithms more respectful.
- recommendation algorithms celebrate their 25th birthday
- Use case #1 : Netflix
- Use case #2 : coupons in the retail sector
- Use case #3 : connected watches and bands in the insurance sector
- Conclusion : there is hope
Believe it or not, recommendations algorithms are not new. They celebrated their 25th birthday in 2017. The first recommendation algorithm, called Tapestry, was developped at Xarox in Palo Alto in 1992. Other early models followed soon after (MovieLens, GroupLens) and the legend says that Amazon got its first algorithm soon after it was founded in 1994.
The first use case I presented belonged to the media sector. I chose the classical Netflix example, well-documented and to which everyone can relate. Next to the general information on what is personalized on the Netflix homepage (including artworks personalization presented at the latest Recsys conference in Vancouver), I also covered some less known aspects on the societal impact of Netflix as Neil Hunt outlined it at the 2014 Recsys conference. This served as a gentle introduction to the topic of filter bubbles.
The second use case I presented related to the retail sector. Following the popular article I published in 2018 on the 5 future trends of the retail sector, I explained how one of the most important marketing activity of retailers (coupons) was going to be impacted. The time variable will soon be captured by retailers who will be able to figure out in which sequence your bought your products and will have the possibility to adapt the in-store journey by sending real-time cues to your mobile devices (or directly on the shelves as enabled by smart shelves). I also explained how, in the near future, retailers will be able to propose coupons when you enter the store rather than when you exit it.
Finally I sketched the future of the insurance sector in the era of data. In particular I showed how smart watches and smart bands will give access to biometric data that will change the way insurances work. The risk sharing principle may be at stake.
One day, I predicted, clients will have to pay for NOT wearing an e-health device or for driving their car themselves.
“computers are useless. They can only give you answers”
As Picasso said “computers are useless. They can only give you answers”. This was the final quote of my presentation which allowed me to give some hope to attendees as far as perspectives on artificial intelligence are concerned. In an article I wrote on creativity and human intelligence, I explained why there is no risk that all humans get replace by robots. We are still superior as far as creativity is concerned. Predicting the future is based on historical data and humans have the amazing ability to disrupt the present and the future, making us irremplacable.Tags: bank and insurance, media, recommendation algorithms, retail