Having been to DTX in London,  I thought I would take a look at DTX in Manchester.  Manchester is a bit of a Tech hub itself.  I remember that when we were looking for Penetration Testing services for PowerSchool in Sacramento, we actually ended up using NCC Out of Manchester.  This conference is at Manchester Central that I assume to be a converted Railway station.  It is a great conference venue.  The conference company seems to try and bring overlapping technical communities together: Digital Transformation, Customer Experience, Cyber Security.  I think it worked quite well, although the customer experience side seemed a bit light.  It was as ever interesting to see who is speaking, who is sponsoring and what themes are dominant.  Maybe no surprise that the AI theme was dominant, although some of the more experienced technology folks there did concede with eagerness that we classify everything as AI these days, so it is becoming a bit meaningless as a term.  These are a few of the sessions that I attended.  I was sorry to have missed all of the Oracle sessions that seemed to have been on the Wednesday.  Oracle was a much more visible sponsor, but I did not see much evidence of Oracle out on the exhibition floor.  

The CEO of Derby City Council was extolling the virtues of their AI Rollout, When he was speaking it was not super clear what his AI Projects were but he was a convincing proponent.  He explained the consultative process they have gone through to ensure good acceptance internally.  He also explained that they had established an AI Ethics Board within the council.  He was on stage with the project leader and a gentleman from OCI.AI who were their systems integrator.  

As it transpired, they have created digital assistants, one for external interactions and one for staff interactions.  These are custom implementations of Microsoft Co-Pilot.  

They claimed significant savings from these chatbots in areas such as Adult Social Care and Debt Recovery.  

The CEO talked at length about using AI to automate the record keeping that plagues the workers in local services, but this was a future project. 

Delivered by a company called Ometis

They laid out a framework for bringing data into an Analytics, Machine Learning and Artificial Intelligence layer.  They are based on Qlik. They intrigued me by talking about conversation as an interface for analytics. They gave some other examples:

• Part failure,  much more ML than what has traditionally been thought of as AI

• Recruitment example, candidate follow-up date and tickler.  This seemed 90% workflow management with maybe 10% AI. 

This was delivered by the Manchester Airport CTO.  He wanted to reinforce the Human aspects of AI.  He was the first of many presentations that talked about ISO 42001.

They are building a chatbot.  Not high tech at all. However, offering more of a NLP interface. Very practical, harvesting the questions as inbound info. 

The other example that came out of the Q&A session, was managing passenger flow as an optimization problem.  Apparently, when the airport waiting area gets crowded, the passengers get anxious and uncomfortable.  So they slow the belt speed at security slightly lengthening the wait time there and they call flights a fraction early.  In the manufacturing world we call this Drum, Buffer, Rope. The theory of constraints being that as the drum beats, you move the work piece into the next buffer and containers are tied by a rope, (at least that is how you visualize it) so as you move one, you move the chain.  I don't think we would have ever thought of it as AI.  It is a very practical solution to a planning optimization problem. I actually spoke to the gentleman after his talk.  I assumed that an Airport Authority, must have had a pretty good handle on queue management for the management of runways, so was surprised that they had chosen an AI/Neural Network approach to the passenger movement problem.  He said that they had learned some unexpected and novel things by running an AI model that they would not have found through a “Procedural” approach.  For example, people driving Audi cars tend to arrive at the airport very early.  

This talk was delivered by BJSS.  It had a feeling of being a bit more technical.  The speaker, talked a lot about augmenting human tasks, rather than replacing human tasks.  He was trying to show that there is a danger that the simple tasks tend to not be considered demonstrations of AI, so the complex creative tasks tend to be done, leaving the simple and mundane to humans.  

He felt that as the benefits accrued to senior people in an organization, it would not be well accepted. 

This was my last session.  To be honest it was too long, and while it did a reasonable review of the different frameworks and standards for AI, the conversation seemed to veer off into the history of ethics in general. The EU AI Act is something that we are all going to have to become familiar with.  A good cheat sheet is here.  A great comparison between frameworks is available here. I actually think it could have been very useful as an ethics workshop.  The examples used would have generated useful debate, but we needed facilitation and a way of recording the utility of what was being done and ethical dilemmas in doing it.