CogX Impressions

In July this year, we attended CogX (short for Cognition X), a three-day event showcasing AI and Emerging Technology, here in London. 

The place was abuzz with exhibitors, scientists, entrepreneurs, captains of industries and investors. Of course, DeepMind, IBM’s Watson and other usual suspects were prominent, as the rightful passengers on the AI (probably autonomous) bus, riding towards a future that has already surpassed some of the science fiction of our childhood.

The event’s underlying theme was that Machine Learning and AI, fuelled by high performance computing and big data, have powers to predict, optimise and improve the status quo. The AI takes a pan-optic, analytic look at the past and gives us a glimpse of a statistically literate future.

I was quite impressed with the variety of services being developed and the increasing number of problems which employ these new applications of mathematics. From fighting money laundering by finding specific patterns in transactions across the globe, to predicting the likelihood of patients turning up for their medical appointments, big data and novel algorithms are combined to extract more informed, expert and long-ranging predictions.

As mathematicians, we find ourselves at the forefront of this innovation gold rush. Torn between scientific territorial ambitions and the impact of science on our future, I hope that our scientific community is mindful of the adage that with big power comes big responsibility.

We are the creators and the gatekeepers of this new power. We are best placed to understand its constraints, limitations, validity and biases. We continually grapple with the data quality, its sparsity or obesity. We are still considering its relationship with the reality or ground truth and our theoretical models.  We are still learning about the challenges of extracting knowledge from data. We also know that sometimes, no matter how big the data is or how hard we try to persuade it, it will still not confess: Unprovability comes to machine learning.

We are just starting to glimpse into a future where AI is our companion.  Events like CogX give us an idea of  what is possible. Some of the people and companies I met at CogX re-framed my vision of this exciting journey. Here are some notable AI pioneers from CogX:

Håvard Haukeland, founder of Spacemaker AI, demonstrated vividly how sheer curiosity and desire to make a positive contribution, took him on a path of achievement and success. His initial idea was to design a high-quality living and sustainable building. In preparation, he talked to people involved in all aspects of what makes a good home, from the local vegetable suppliers to light optimisation experts. He realised that nobody had combined all this valuable know-how with data to help architects simulate their creation.

Three years later, his newly created program for building optimisation is used by over thirty real estate developers and architecture firms in Europe. If there is a start-up land, this is a perfect fairy tale from it.  Håvard’s story strengthened my conviction that the most fulfilling achievements spring from one’s core values.

Tom Wilson VP Automotive of Graphcore, shared a personal insight about teaching AI how to drive. A driving instructor in his early career, Tom was musing over the concept of “how to teach cars to drive”.  The questions of “how do we know what we know” and “how do we program it into computers” is very pertinent in the world of AI and leads us into the sea of epistemology. And short of jumping in with ancient Greek philosophers, I will quote Tom:

 “I am beginning to think that more autonomous vehicle start-ups should employ ex-driving instructors. To a certain extent, there is a part of driving that has nothing to do with building a 3D map of the environment.  Much of what we teach students is about driver psychology and gauging driver intent through body language, making eye contact, etc. No LiDAR tells you that stuff. Driving instructors do understand driving in a way that AI engineers working on autonomous vehicles (many who don’t drive themselves by the way) don’t!”

And last but not least:

Dimi Masaouti CPO & Co-founder of WinningMinds AI showed me how their application is able to analyse conversations and reveal team dynamics.  Drawing on behavioural science and social psychology as well as mathematical analysis, WinningMinds designed metrics to describe team effectiveness, collective orientation, decision-making, levels of agreement and problem-solving focus within a team.

So the next time you are attending a long meeting and would like to reach a solution faster check out WinningMinds AI.

If you are interested in exploring how Machine learning and AI may help your business please get in touch.   

Leave a Reply