Once again, GDR attended the annual Artificial Intelligence in Business & Entrepreneurship (AIBE) Summit keen to find out about all the latest developments in AI. A recent Justin Timberlake music video had somewhat corrupted our expectations of events like these, but despite the absence of dancing androids we were saved from dejection by a stellar line up of speakers and exhibitors offering valuable insight into progress being made in the field.
The event uses a wide lens to examine AI’s implications for business across various industries, from defence to agriculture, yet throughout the day we were able to learn much about developments in AI that represent exciting opportunities for retail.
Here are the things that really stood out for us on the day:
Real-time video analysis is big news for stores
Amazon Web Services employs thousands of engineers focused on deep learning, and the company’s Technical Evangelist Danilo Pocchia spoke at the conference about his team’s latest projects. A significant expansion to Amazon Rekognition, AWS’ image analysis tool, has recently been rolled out, which means the technology is now able to apply the same analysis to video files and even live streams in real-time. This means that in video footage Amazon Rekognition can identify, not only objects and faces, but also actions.
There are very exciting potential applications of this technology in retail, from security and seamless loyalty programmes to detecting areas of the shop floor that need replenishing. In April, Amazon will launch Deeplens; a camera with similar software built into the device, designed for developers of all skill levels to enhance their deep learning programming abilities and develop their own AI-driven applications. It is likely then that the number of use cases of real-time video analysis in retail will begin to proliferate.
AI in marketing isn’t just about chatbots
An exhibitor that really grabbed our eyes (and ears) was a startup called AIVA Technologies, whose AI uses deep learning to generate soundtracks for entertainment content such as commercials, video games and films. Aiva analyses vast quantities of classical content, and can then compose its own track to match various tones and emotions. Aiva’s music has already featured in a commercial for Vodafone and raises interesting questions about the future role of AI in the creative industries.
AI also allows greater efficiency in analysing social media activity. Poccia spoke of AWS’ work in developing multilingual social analytics, which could radically reduce the time spent gauging engagement across different markets, or indeed in a single area where multiple languages are spoken. These new tools therefore promise centralised social analyses where currently multiple teams or platforms are necessary.
Things to bear in mind
In amongst the showcasing of the AI capabilities of the various companies exhibiting or speaking at the event, there was also some sound advice and food for thought in terms of how AI should be applied to business.
Hugo Pinto, Managing Director at Accenture Digital, offered some practical advice. In consulting businesses across various industries about AI implementation, he has discerned a common pattern of companies either not collecting enough data, or not collecting it effectively. Too much offline data is not recorded at all, and in many cases data that is digitally recorded is done so in siloes, which hampers the efficacy of putting it all to use in AI. Pinto provided a factory in a manufacturing company as anecdotal evidence, but it is easy to see how this can also apply to stores. In short, as much data as possible should be recorded and stored in one place in order to provide the food that AI requires to produce the best outcomes.
Then came some broader warnings with regards to ethics from Professor Luciano Floridi, Director of Digital Ethics at the University of Oxford. He quickly dismissed the notion of the evil, sentient AI that we see in films and written about in the media, but said that the challenge is instead about digital governance, not the innovation itself.
Floridi cited a range of ethical implications that follow what he terms the divorce of intelligence from agency; in other words the ability of machines to achieve human results without their human intelligence. Among these were enabling human wrongdoing in terms of future misuse of AI technology by organised crime and cybercriminals; increased human dependency on digital and a reduction in human accountability.
But most relevant for individual businesses is the current opportunity cost of implementing a technology for which very few rules presently exist. A lack of regulation means that at this early stage there is considerable potential for PR disasters if consumers perceive an abuse of its capabilities.
All in all then, a fascinating day of discovery and insight. There is still a sense that it is early days in the field, but the proliferation of startups, breakthroughs and tangible applications to retail leaves little doubt that AI will play an ever-growing role in the industry.
For more information on the AIBE, visit its website.