Key takeaways from Access AI briefing with Google Cloud, Lloyds Banking, Cognitive Finance, and BIMA

Posted: 23 November 2017 | By Charlie Moloney

This morning it was our great pleasure to introduce speakers from Google Cloud, Lloyds Banking, Cognitive Finance, and BIMA, to the stage at an Access AI Breakfast Briefing event in Spitalfields, London.

The panel included:

For those of you who couldn’t make it today, we’ve distilled the key takeaways from this event:

  • What is AI?: The panel agreed that at the moment the term artificial intelligence (AI) and what it means is determined by the task you are trying to achieve. Instead of thinking about generally intelligent systems, businesses need to identify pain points and then think about how intelligent tech can be applied there. Yuval Dvir pointed out that when we think about intelligence, we think of something similar to human intelligence, and he said this is not AI, which is currently being used in a narrow way to solve specific problems.
  • Change begins at the board: Something that Clara highlighted is the fact that you have to get the C-suite of your company to accept and push the implementation of AI if you want to see results. It’s not something that you can just leave in the IT centre. The board needs to be on it. Top down implementation. They need to be intellectually curious about this tech, as she mentioned she was impressed with a C-suite exec who registered for a Python course.
  • Culture and transformation change: Pete and Chris discussed what they had learned working on an AI project together for finance: too often the reason for and ultimate goal of implementing AI is not understood. Cultural shift needs to focus on humans at all levels as the driver of this change, rather than just technology.
  • Looking at the long term benefits: An important area discussed was the overemphasis on short term return on investment (ROI). ROI is too often the thing companies look for when they invest in AI, but really they need to be thinking of a much longer term strategy. In the short term it might mean expenses in terms of retraining, teething problems. In the long term you won’t be able to compete without AI.
  • Transparency: There is a difficulty emerging with so called ‘black box’ technology, i.e. systems where you can’t understand why it made the decisions it made. The alternative open systems, some people call them ‘white box’ technologies, where you can clearly understand why it’s made the decisions its made, are still a problem because you’ll need someone to work full time on reading it, looking for bugs, and looking for insights. That’s why Chris Popple suggested creating AI systems that can check that other AI systems are working properly, but this is technology that requires more development.
  • Ethics + Responsible AI: The audience queried the pros and cons of open algorithms, who controls data, and diversity. Some specific matters discussed were Facebook’s algorithms being made public for anyone to use, the AI system that was created to judge women’s beauty, and who owns the knowledge that machine learning systems learn.

We greatly appreciated the speakers contributing their expertise to the event, and we want to thank everyone in the audience who came and contributed questions and idea.

We hope to see you at our next events in 2018!

Related industries

, , ,

Related functions

Related organisations