My report from TechEx London.  I spent the day back at London Olympia.   NHS Supply Chain had myself and one other delegate.  TechEx has merged a few technology themes to gather an audience whose interests will certainly overlap. 

• AI and Big Data

• Block Chain

• Cyber Security and Cloud

• Digital Transformation 

• Intelligent Automation

• IOT Tech

This was a very soft skill presentation, but I was very interested in the claims about getting executives to think about concrete examples of data value.. I have generally found execs to be convinced in abstract but it is difficult to get people to give examples. 

I thought Jason’s understanding of system 1 and system 2 decision making in humans was interesting.  You can see the overlap between fields and people talking about machine learning start quoting psychologists.  

His conclusion was that humans should trust the results of the analysis more and their instincts less.  I am not sure that is true.  I would say if the analysis does not conform to your instincts, check the data.

This was a fascinating panel discussion among real practitioners in AI.  They have all been working in the field, long before it was fashionable.  They gave some good examples of where large language models can provide value.  

One example that resonated with myself was creating synopsis for service request trails and bugs.  Thorny issues can have bug text that has been added to over many weeks by many people.  An engineer picking up the issue has to read through all the context.  

When asked if the buzz around AI is real utility or just hype, they leaned toward hype.  

What was even more interesting was when asked if the dangers were real they leaned toward real.  

When asked what we should expect from AI in the next year, they said that fake linkedin requests will overwhelm us.  I thought that was a profound insight from Meta, Ebay and Bumble. 

They also thought new laws, regulations and standards were both crucial and imminent.

They thought the word predictor capability that has yielded ChatGPT, is unlikely to yield Artificial General Intelligence.  They thought the route to AGI is not through large language models and that we will have to start again. 

I thought this session would yield some profound insights given my current role with NHS.  However, it seemed to be a bit lost as a session.  There seemed to be some insights that using apps for mental health tracking can give broader coverage in service provision. The presenters seemed very pro technology and very pro social media, I think that was worthy of a much more in depth discussion.  There will be a replacement of clinician to patient interaction with apps.  There are costs and benefits to this that need to be carefully weighed.  There was some conversation around using social media posting as predictors of mental health crisis and the privacy issues that surround it.  There was frustration raised on how difficult technology companies are to work with for clinicians.  I can see that smaller companies can give lots of attention.  I also see the instinct within the NHS Supply Chain to talk to smaller companies adding apps into the ecosystem rather than ask the larger companies how their applications work.  It leaves lots of capabilities in the larger systems unused, systems less coherent and data more isolated.