Stephen Wilkinson

I'm doing a lot of study and learning on AI at the moment and I'm noticing something that's really irritating me - and it's the mental model people are using to discuss it. (including course teachers and designers.)   The idea that an LLM can be "confused", or "it didn't understand", or "ChatGPT told me…." shows that we are ascribing some very human attributes to what is essentially a probability machine.   While people talk about the ephemeral algorithm targeting them or others with social media content, the mental model used to discuss AI is very different. It is discussed as a person rather than a tool. If it were discussed as a tool people would say "the GPT output was…." "the output was incorrect, I had to prompt it differently to get an accurate output". These models are trained on the basis of how well they please the user. The unstated assumption here is that the user will be most pleased by accurate data and information. But really - there is a volume of users that are most pleased by content that APPEARS to be true and accurate. Good enough is good enough. People use generative AI as a convenience tool so there is a proportion (I suspect large portion) that will not make the time to verify the outputs.   And interestingly, the more we view AI as a person, the more likely we are to blame it, as though it were the tool, not the inputs or our participation that was the problem.   We've passed the uncanny valley and here we are about to fall into some very avoidable traps as the language we use to talk about LLM outputs creates a cultural misunderstanding.

207

111 Comments

Ever sat in a room with 20 senior people and watched an hour disappear… with nothing decided except all the reasons not to move forward? I was in a meeting recently with 20 senior colleagues. The purpose was simple: discuss how we could get a trial of something new that we really need to get a grip with up and running quickly. What actually happened? We spent an hour hearing why we couldn’t move forward. Budget. Risk. Processes. Governance. Uncertainty. Fear. One by one, the reasons stacked up. The energy in the room was focused almost entirely on why progress wasn’t possible. This isn’t unique to that meeting. In higher education, we’re brilliant at picking apart the risks and obstacles. But if all the energy goes into “why not,” there’s none left for “how can we?” But – it is us in those meetings. We can change it. Sometimes you just have to find a way to make progress happen. The mindset shift is simple but powerful: From “we don’t have enough resource” → “what’s the smallest version of this we could trial?” From “we’re not ready” → “which bit is ready to go first?” From “what if it fails?” → “what will we learn if it does?” Real progress comes from trying, learning, and iterating - not from waiting until every possible objection is resolved. The next time you find yourself in a room full of reasons not to act, maybe the best question to ask is: “what would it take to just get started?” Skip the hour of objections - start out with accepting them (without listing them all out). Change the tone to 'This is tough, with a lot of reasons to not do it. But we must. So - how do we move forwards? What things can we do?' How have you managed to flip that energy in your institutions from “we can’t” to “let’s try”?

74

13 Comments

Learning styles have changed. We have lots of learning opportunities in DWP Digital Delivery, but one of the most popular is the 15 minute knowledge blitz. These fortnightly bitesized sessions are about topics which attendees should know (but we can't all know everything) delivered by experts in that thing. We've covered topics like complexity, engagement skills, Wardley mapping, DORA, facilitation, team topologies and many more. The recordings have built up into a library even more popular than live attendance. So far, we've had over 1,000 hours of focused, relevant learning. Not we're opening them up cross-government. If you're a Digital Delivery person in government, you're welcome to attend, and we're working on making the library available. If you're in another organisation, this is a pattern you're welcome to copy - it's free and highly effective. The last session was the ever brilliant Mike Bainbridge on productivity, as is the next one.

69

11 Comments

I'm thinking of starting a few regular things I did for years at Sage. The first is a regular 'intro to accessibility' webinar for people that could do with an introduction to digital accessibility. Probably 25 minutes weekly and it'd be the same presentation every week, with a short discussion afterwards. It'd be for product owner/manager type people, as well as developers, designers, QAs, and so on that want a brush-up on the very basics: 🤔 What digital accessibility is ❓ Why it's important 🛠️ A couple of easy to understand examples of issues and fixes 📚 Some resources to check out What do you reckon?

65

11 Comments

Starting today, I’ll be sharing a new digital course every day this week on the QA Ltd platform. Each post will focus on either #multicloud strategies or #cloud #costoptimisation, giving you practical insights to boost your cloud #skills. 🌐 Day 1: Multi-cloud Architecture Foundations https://lnkd.in/ebpjHGnD Kick off the week by exploring the essentials of multi-cloud architecture. Today’s course will help you: ✅ Understand the fundamentals of multi-cloud architecture ✅ Analyze common use cases across different cloud providers ✅ Evaluate the benefits and challenges of multi-cloud environments ✅ Apply design principles and architectural patterns to real-world scenarios Whether you’re starting your multi-cloud journey or looking to refine your architecture strategy, this course sets the foundation for building resilient, scalable, and flexible cloud solutions. Stay tuned and don’t miss a day! #MultiCloud #CloudArchitecture #CloudStrategy #FinOps #DigitalLearning QA North America Thomas Mitchell #Cloud #Learn #AWS #Azure

34

1 Comment

We talk a lot about accessibility compliance and conformance, but accessibility isn’t just about ticking boxes or meeting standards. At it's core accessibility is about enabling people. We're aiming to create digital experiences that allow everyone to participate independently and equitably. For disabled people, accessibility is a fundamental necessity. It removes barriers, enabling access to information, services, and tools many of us take for granted. But the impact of accessibility doesn’t stop there, it ripples out to benefit a much larger audience. We're also improving things for people as they get older. Things we took for granted, like our mobility or vision, degrade. And we might no longer be able to mentally comprehend fancy new interfaces. Accessibility also helps out in situations where you might have a temporary disability, like a broken arm or eye infection. Or you're in a practical situation where you're holding something and only have one hand available, or are on the bus without your headphones and those subtitles turn out to be really useful. Microsoft's Inclusive Design information does a good job of explaining these concepts. Consider this wider effect when talking and thinking about accessibility. It'll give you a better end product that everyone can actually use. And that's what we aim for, right? #Accessibility #A11y #Inclusion #InclusiveDesign #DesignForEveryone #RonnyTalksA11y #Toegankelijkheid

44

2 Comments

Had a few conversations lately with local gov folk which bemoaned the lack of active user groups for most software systems. This is a problem! What could we do about it? It seems like vendors are keen to say they have user groups, but then once the sale is made, less keen on convening them. There's real advantages in the people actually using software to get together to share insight, issues, collectivise around requests, and so on. Is there a space, I wonder, for an independent user group as a service offering? Someone to provide a safe place for online discussions, organise regular meet-ups, do a bit of the admin, and maybe engage with the vendors to get them to turn up and so on. Question is probably 'who pays?'.

24

37 Comments

The Hasso Plattner Institute has produced some of the best free online training for Green IT that I have come across so far, especially their 'Sustainable Software Engineering' course (see https://lnkd.in/ef_-hHGA). I am looking forward to starting their 2 new courses, especially 'Environmental Impacts of AI Systems' (see https://lnkd.in/e28_ZJKW) which starts March 4th, but also 'Energy-Efficient Software Development' (see https://lnkd.in/egCuRh3X) which starts March 25th. Each course includes the option to take an exam as well as the option to just participate. Highly recommended!

48

9 Comments

💡 I recall @Matthew Skelton once describing HR as the “accidental architects of software systems” in a presentation from a few years ago. This is because the way HR defines roles, teams, and career paths often dictates the structure of software delivery, whether or not it’s intentional. The BBC just covered a trend where HR and IT leadership are being merged, driven by AI’s role in reshaping how work gets done. In theory, this could turn accidental architecture into intentional architecture, aligning people decisions and technology decisions in one place. So my question is: Would merging HR and IT leadership help design better socio-technical systems, or would it merely create a larger silo with two distinct mindsets within? https://lnkd.in/e9S_AyWV #sociotechnical #orgdesign #futureofwork #AI #teamtopologies

34

7 Comments

Why do clients engage with Yorkshire Software Solutions Ltd and myself...? Because of trust! I've spoken a lot about trust before, but it really is as simple and pivotal as that. - They trust we will tell them the truth (whether good or bad) - They trust we will deliver quality - They trust we will get the job done Obviously building trust is a skill based, which is largely based on outcomes. You have to live and breathe a culture of building this trust. Some small things that will help you build truth with a client: - Don't go in and say everything is terrible (Common practise for some contractors I know) By highlight the positives aspect of their software and architecture, as well as identifying what needs work, it creates the feeling that you can be objective and are not just looking at how much money / billable time you rinse them for - Give honest estimations Nothing will alienate clients more than giving estimations that continually get pushed back. Do the ground work, set proper expectations, and IF it looks like the estimations were off, let the client know asap, providing them with options. E.g. How can we get back on track? Do we want to descope for MVP with fast follows for release date, or delay the release? - Provide advice whether it benefits you personally or not Doing this shows your professional skills in the industry. If you provide advice for instance that saves the company time and money, but as a result, means you have less chargeable work, the company will trust your opinions, so when something big DOES need doing, they know who to call. This is not everything, but just a few suggestions from my experience on how to build trust with clients, stakeholders and even your line manager and team. By using these techniques, every major client of Yorkshire Software Solutions Ltd's has had repeat engagements with us. What are your techniques for building trust? #Trust #Honesty #Software #ComputerProgramming #SoftwareTeams #Leadership #HappyClients #SoftwareDelivery #RepeatEngagements #HighQuality #Estimations #GetttingShitDone

20

3 Comments

Guest lectures are useful. Company visits are nice. But that's not the same as students working on problems organisations haven't solved yet. This year, my Strategic HRM module is partnering with UK Health Security Agency on a real challenge they're facing. A few weeks ago, Lisa C. and Ben Rush, came to campus to share a strategic workforce development priority they're exploring. Not a polished case study about what worked, but a genuine opportunity for rigorous analysis from outside their organisational context. The students asked good questions. The kind that come from reading performance management theory and change management research, then realising those frameworks get complicated when you're dealing with actual resource constraints and competing priorities. Now they're developing #strategic recommendations they'll present back to UK Health Security Agency in December. Not essays for me to mark. Strategic analysis for the people who might actually use it. That's the difference between teaching HRM and #preparing people to do it. Students can't hide behind textbook answers when they know real #stakeholders will challenge their thinking. They learn that strategic thinking means navigating complexity, not reciting frameworks. For UKHSA, they gain an external perspective grounded in current academic research - a different lens that complements their internal expertise. Fresh analytical approaches matter when tackling strategic priorities. 👉 Genuinely grateful to the UKHSA team, Lisa C., Ben Rush and Mark Elliott, for reaching out, investing their time, and trusting this approach. These partnerships only work when organisations are confident enough to invite external scrutiny and value fresh thinking. For other organisations thinking about university engagement: this matters for both sides. You get early connection to people who'll shape your profession, and we keep teaching grounded in what's actually happening. If you're interested in exploring this kind of partnership, the conversation's open. #StrategicHRM #BusinessEducation #IndustryEngagement

42

1 Comment