Scottish Web Folk 2026: Humans vs Machine
There is no doubt that there has been a step change in how Higher Education has delivered web services over the past few years. However, stepping foot in the University of Dundee’s conference hall for another Scottish Web Folk, I’m delighted to report that it’s still very possible to fill a room with brilliant people who are happy to nerd out about our work!
You can read about my previous experience here:
Scottish Web Folk 2025: Challenging times, brilliant people
Old rules no longer apply
As is tradition, the University of Dundee’s Andrew Millar, opened the conference with a presentation which presented the theme of the day: “Human vs Machine”.
He put forward the idea that the amazing technologies which are transforming digital services also come with significant risks and challenges.
For example, indicators which were once used to assess the effectiveness of a web service can no longer be relied upon: Page views are less relevant now that many users will find an answer from a chat bot without visiting a website. A site’s ranking in search engines is less relevant now that a chat bot can find answers from anywhere on the web. And so on.
There is tension between the perception of human- and AI-generated content
While there is undoubtedly an increase in the use of chat bots to find information, a recent Gallup survey of 14-29 year olds found that only 3% trust work solely made by AI, 28% trust AI-assisted work, and 69% trust work made without AI.
![]()
In a panel discussion, Marta Krukowska from the University of Bangor described a situation in which an image purchased from a major vendor and used on their website turned out to be AI-generated. This caused some reputational damage in the student community as it put the doubt in users’ minds that other content might also be AI-generated.
The perception of these technologies is clearly in flux and are likely to shift significantly over the coming years. While it may be beyond us to influence this process, Millar put forward what he sees as our role:
- Advocate: Speak for the user who is not in the room through research, accessibility, and plain language.
- Defend: Protect accuracy and our voice. Ensure that what AI says about us is correct. Guard our data and intellectual property.
- Decide: Own what is published in our name through approval thresholds, escalation, and a working stop button.
Digital practitioners have moved from theory to practice
Ensuring that our web content is correct and parsable by AI tools is essential going forward.
Only thinking about human users can have unintended results
Rob Ryder-Richardson from the University of Dundee described a situation in which something that was designed to help human users had unintended consequences when interfaced with AI.
Their course information pages dynamically displayed different data depending on the geographical location of the visitor – good for humans, but AI crawlers are almost exclusively based in the United States.
Therefore, when asked about course entry requirements, some chatbots were confidently asserting that the University of Dundee only accepts US SATs and that UK-based qualifications were ineligible.
The solution was to essentially develop a dataset which AI crawlers could use to find information, invisible to human users who would likely find such information unnecessary and overwhelming – structured in a way which is clear and unambiguous.
AI can be used proactively to improve the quality of web services
Richard Saunders from Zengenti described the benefits of developing a collaborative toolkit which team members can use to assess their webpages proactively to improve quality.
As a manager of a service which contains a surprising amount of content, one of the biggest challenges I have is ensuring that this content is accurate, pertinent, and accessible. While this task is currently done manually, there is certainly benefit in automating this task with AI skills.
In contrast with using one-off prompts, AI skills are plaintext workflows which provide a consistent instruction to the LLM. These work best when each one is focused on one particular outcome, and so developing a library of these within a team can provide a toolkit which can be used to consistently audit a website’s content.
I look forward to applying new best practice
While the technical implications for AI often dominate the conversation, Scottish Web Folk reminded me that successful digital transformation is fundamentally a human endeavour. Two sessions from the University of Edinburgh highlighted to me that before we can deploy such technologies, we must first master collaboration.
Neil Allison presented a masterclass on delivering effective workshops. Too often, digital teams conflate “meetings” and “workshops” and gather stakeholders only to disseminate information rather than create it. He outlined common pitfalls that cause a workshop to be ineffective: such as a lack of clear objectives, prominent voices stifling creativity, and lack of tangible outputs.
![]()
Lauren Tormey delivered a case study on developing content design playbooks. Tormey explained how to co-create a set of best practices using a repeatable, structured workshop format.
As part of a team who hopes to deliver significant changes to our digital services in the near future, I found this talk to be invaluable and look forward to employing these strategies in my workshops soon.
We have a responsibility to help deliver better, future-proofed web services
AI is no longer a distant, theoretical concept, but an immediate operational reality. The path forward requires focusing on both delivering effective technology and ensuring excellent collaboration on the human level.
To take this forward, I propose the following action plan:
- Ensure our success metrics are defined according to user requirements.
- Audit our content for AI readability.
- Deliver excellent workshops to determine our course of action.
Or in other words:
Advocate. Defend. Decide.