A day out at the Public Sector Data Science Summit

I was lucky enough to spend yesterday at the Public Sector Data Science Summit in London. It turned out to be a really good day — lots of great talks and really interesting content. I live-tweeted the day (is that still a thing?) and my tweets are interspersed in the summary below:

  • looking forward to today’s sessions at the Public Sector Data Science Summit in London #psdata18

The first and second sessions, chaired by Lord Clement-Jones, covered updates from The Office for Artificial Intelligence, the Robotic Process Automation Unit of Cabinet Office and the London Office for Technology & Innovation at the GLA:

  • the Office for AI is a joint initiative of @beisgovuk and @DCMS — remember that D is for Digital these days #psdata18
  • the Office for AI has five key areas… leadership, skills, data, sectoral adoption and missions #psdata18
  • “international competition for people with AI skills is very strong” says @StephenHennigan of the Office for AI — looking to increase number of PHDs in this area and attract skills from overseas #psdata18
  • “we are working with the ODI to create a series of data trusts” — says @StephenHennigan of the Office for AI at — also working closely with @GDSTeam to “review and drive adoption of AI across public sector” — but public trust in AI is critical #psdata18
  • “the Office for AI is new, so it still needs to establish itself as the go-to place for information and guidance about the use of AI in the public sector” says @StephenHennigan at #psdata18
  • dunno if I’m missing the obvious… but I can’t see a Twitter handle for the Office for AI — odd? maybe they are waiting for it to create itself?? :-) #psdata18
  • RT @AdaPeck: Have to bring public with us on this says @StephenHennigan talking about new Centre for Data Ethics & Innovation . That’s certainly a challenge but vital imo especially in #health #care #PSDATA18 @PSDataScience
  • “robotic process automation (RPA) has huge potential for improving the delivery of public services in the public sector” says James Merrick-Potter of the @cabinetofficeuk at #psdata18 — @HMRCgovuk and @DWP cited as examples of how this is working at massive scale
  • big gap between @HMRCdigital, @DWP and everyone else in the RPA space — hence the creation of the Centre of Excellence for RPA — partnership between @cabinetofficeuk and @Capgemini including delivery of services that build RPA solutions for government departments #psdata18
  • RT @sprague_daniel: AI is predicted to increase UK GDP by 10% by 2030, according to the Office for Artificial Intelligence #psdata18
  • some RPA examples… @HMRCdigital have cut call times by 40% using RPA — @DWP have reduced manual effort producing PIP Appeals Bundles by 35% — @educationgovuk have automated the logging of emails into their correspondence handling system #psdata18
  • lessons learned… get the right people on board, look beyond the PoC for funding and building a business case, choose the right processes to automate, build confidence, start small but think big — says James Merrick-Potter #psdata18
  • “automation will play a key part in the transformation of the civil service” says James Merrick-Potter of the RPA Centre for Excellence #psdata18
  • “although we are engaging with 40 departments, mostly it is at the level of individual projects — a wider review is required to understand the strategic impact of RPA across whole departments” says James Merrick-Potter of the RPA Centre for Excellence at #psdata18
  • @hopkinsdavid Yes, ethics definitely a focus of activity for the Office for AI #psdata18
  • London Office of Technology and Innovation #psdata18 https://t.co/3QQ0ubSBcC
  • “working across multiple independent London Boroughs is one of the key challenges to AI adoption (and smart technology adoption more generally) across London” — now multiply that up by every local government org in the UK! there’s the real challenge for the UK… #psdata18
  • Strike a New Deal for City Data — https://t.co/AQzvKtA6KE #psdata18
  • “there will never be a single data lake for all public sector data in London” says Paul Hodgson of @LDN_gov “but we do need there to be access mechanisms and technical linkages between appropriate subsets of data held across London” #psdata18
  • RT @David_Miller18: “There; will never be a single warehouse for all of London’s data” — Paul Hodgson. We need to connect systems so we can share the right information and metadata when appropriate. #psdata18 https://t.co/eyONY4QZsF
  • “everything we do [with public sector data] needs to be legal but I think we need to set ethical standards that are higher than that” says Paul Hodgson of @LDN_gov at #psdata18
  • RT @DrCodiroli: Fascinating overview of the use of data and AI in national & local government this morning at #psdata18. Ethical standards and knowledge sharing integral.
  • Q: what is the impact of RPA and AI on jobs? (someone had to ask!) A1: “RPA will help local gov do more with less (and they are already have less due to cuts!)” — A2 “automation isn’t taking jobs, it’s taking parts of jobs (think tasks rather than roles)” #psdata18
  • RT @DamiAwobajo: Argument at the #psdata18 about whether automation is taking jobs. View from a panel member is that “we are automating tasks not roles”.
  • RT @Clc_Schreiber: Interesting morning so far hearing about the government initiatives in improving services through tech, innovation and AI. Shame it was an all-white-male panel though. Diversity ouch… @PSDataScience #psdata18

Following the break, the second session was made up of talks by Sir Anthony Seldon of the University of Buckingham, Tom Smith of the ONS and Professor Helen Margetts of the Alan Turing Institute:

  • “the most important thing we can all do at the moment [in this AI/AR/VR/digital/social media world] is to remember what it means to be human” — Sir Anthony Seldon @AnthonySeldon of @UniOfBuckingham speaking at #psdata18
  • how will AI transform education… “many of you will have been at school yourselves” :-) — @AnthonySeldon speaking at #psdata18
  • “is AI opening our eyes, shielding them or blinding us?” — @AnthonySeldon speaking at #psdata18
  • RT @EileenLogie: @Clc_Schreiber @andypowe11 @PSDataScience Loads of innovation and AI stuff going on at @ESFAdigital and @DfEDigital in a great diverse environment!
  • five endemic problems with education (the factory system) today… 1 lack of social mobility, 2 one size fits all, 3 teachers drowning in admin, 4 focus on narrow cognitive ability, 5 homogenisation rather than individuality — AI will/should help in all cases #psdata18
  • “education is currently preparing young people brilliantly for… the 20th century” — @AnthonySeldon speaking at #psdata18

After suggesting that if the public sector doesn’t step up to make education fit for the 21st century then the private sector certainly will, Anthony Seldon went on to say:

  • “people who work for the big tech providers aren’t bad people but they are driven by forces that are not necessarily in the best interests of ordinary people and our towns and cities” — @AnthonySeldon speaking at #psdata18

My thoughts…

  • interesting/ironic that we are here to talk about data and AI and yet one of the major failings of UK government around the provision of education at all levels is that they haven’t taken any notice of what the data is telling them #psdata18

At this point I realised I’d been tweeting Anthony Seldon’s name wrongly. Bit embarrassing (and I have corrected it in the above but still a bit embarrassing!). Note that in order to illustrate the failing of the IQ measure of intelligence, Anthony Seldon had partitioned the audience into three equal parts — the intelligent people, the average people, and the dullards.

  • Apologies… Anthony Seldon (not Sheldon) :-( — @AnthonySeldon — in my defence… I was sitting in the section of the audience labelled as idiots #psdata18

Cough. Back to the ONS:

  • building world-leading expertise in the innovative application of data science — the @ONS Data Science Campus — see https://t.co/aHyYuDYjOa #psdata18
  • the @DataSciCampus Data Science Accelerator — applying data science to practical and real challenges in the UK public sector — https://t.co/1etWbG8yut #psdata18
  • Alan Turing Institute — the national institute for data science and artificial intelligence — https://t.co/VPV6dNWeW2 @turinginst — exploring how data science and AI can inform gov policy, build ethical foundations, improve services and contribute to AI policy #psdata18

Lunch break.

The first part of the afternoon was a panel session chaired by Eddie Copeland (Nesta’s Director of Government Innovation) and including Emma Presley (DWP), Tim East (HMRC) and Andy Wall (ONS):

  • based on a quick show of hands, the vast majority of the room at #psdata18 say they are data scientists working in the public sector — I’m surprised… only because I didn’t realise there were so many data scientists out there!

Probably just me!

  • “data science is not just about playing with data — it’s about understanding the wider context within which data sits and the ethical issues about how that data should be used” — @AndyWallONS speaking at #psdata18
  • “data science is not the most glamorous of professions” — @AndyWallONS tries to win over the audience :-) #psdata18
  • as a data scientist in the private sector “I got sick and tired of helping people to sell more sausages and jumpers” so I joined the civil service instead… but salary levels are a problem for recruitment — @EmmaPresley at #psdata18
  • well done to @EmmaPresley for pointing out that the make up of today’s speakers does not reflect the breadth of the audience in the room — mind you, that’s probably still true of nearly all conferences #psdata18
  • …more broadly, “what can each of us do to point out things that look wrong from an inclusion/diversity perspective?” says @EmmaPresley — change, if we want it, rests with each of us #psdata18

Despite these comments, I have to say that I was surprised at the level of diversity in the audience. Coupled with quite a large proportion of (how should I say this) younger practitioners, which was great to see.

The day ended with three breakout sessions, of which I chose to go to the one on Data Analytics, including talks by Mike Molloy (HMRC), Katie Davidson (DHSC) and Neil Crump (Worcestershire Office for Data Analytics):

  • “in @HMRCgovuk, the most creative ideas happen when we bring together data scientists with people who have an investigative background” — Mike Molloy speaking at #psdata18

Referring to Matt Hancock, Katie Davidson quipped:

  • one shortcut to improving adoption of AI with a government department is to “get a secretary of state who is passionate about AI and tech” :-) — @KatieDa7a5 of @DHSCgovuk speaking at #psdata18

But she then returned to a more serious note:

  • “Google searches for chest infection and pleurisy correlate really well with hospital admissions for pneumonia but peaks about a week before — so can be used to make near term predictions for hospital admissions” — @KatieDa7a5 of @DHSCgovuk at #psdata18
  • “we been running a ‘Data Science Season’ at @DHSCgovuk to increase buy in by senior management — including activities we were going to do anyway but branded better” — delivered beginner sessions about R, an R Coding Club and seminars — @KatieDa7a5 at #psdata18
  • “we also ran a 14 week cloud-based PoC programme — encouraging staff to experiment with data science — 2–3 projects have gone live as a result, with associated director buy-in and funding” — @KatieDa7a5 of @DHSCgovuk at #psdata18
  • “data science is not something that data scientists do in isolation — we need the involvement of the wider team” — @KatieDa7a5 of @DHSCgovuk at #psdata18

Finally, Neil Crump spoke about the work being done in Worcestershire to increase data sharing and use:

  • “there are fundamental data sharing building blocks that need to be in place in the context of local government before you can even begin to think about the value of data science” — @neillcrump of WODA speaking at #psdata18 — https://t.co/YTDV6g4npD
  • “we’re not that good at being clear about, and measuring, the value of using data to improve services” — @neillcrump speaking at #psdata18
  • “in a multi-agency environment, there will already be people doing data science — the challenge is getting them to talk to each other” — @neillcrump at #psdata18

If there was anything disappointing about the day it was the number of talks that were basically, “we’re a new government body, we’ve been created to do X (where X was some aspect of data science or AI), and here’s an outline of what we propose to do about it”. I was hoping for a little more practical experience (which really only got covered in the final sessions).

There was quite a lot of discussion about the ethics of data and AI and about the need to create a meaningful ‘data science’ career path within the civil service (coupled with the challenges of competing for staff with the private sector). This latter topic, I imagine, echoed similar debates around digital in the early days of GDS.

Finally, there was a recurring theme about the fact that data scientists do not exist (or work) in isolation and that the best / most exciting / most innovative results happen when you mix the skills of data scientists with those of others in the organisation.

None of that is intended to be critical. Overall it was a really interesting day:

  • really good day at #psdata18 — loads of interesting presentations — over and out!

Final thought… I was surprised at the very limited use of Twitter as a back-channel during the event. I doubt that there were more than 10 of us tweeting — possibly fewer. Maybe Twitter has fallen out of favour? Clearly, this is a significant and pretty vibrant community of data scientists and related professions. Yet, there was almost no evidence of any level of collaboration or community from the Twitter hashtag. Yet I guess these people must be talking to each other in some way. I’m genuinely interested if that is happening and, if so, how?

There’s no doubt that data, data science, AI and ML are flavour of the month right now. Probably flavour of the year. Probably flavour of the next 5 years! There was a big crowd for this event and lots of evidence of ambition to use these technologies to transform the public sector, and the services it delivers, for the better. Great stuff.