Skip to main content

  • 21 May 2021
  • by Andrew Girdwood

Scotland's AI strategy and what it means for your business

Andrew Girdwood unpicks Scotland's AI Strategy and what this means for businesses.

I use artificial intelligence (AI) every day here at Leith, like many other digital marketers. It is an essential and entirely non-glamorous part of business as usual.


Now, I know it's become a point of contention for any business to claim that they use AI, so I'll get back to that in just a bit. I first want to talk about Scotland's AI Strategy, which came out in March, and which I followed closely.

Scotland's AI Strategy

As any good strategy review does, Scotland's AI Strategy moves quickly to describes the subject matter. Here's how the document defines AI.

"Technologies used to allow computers to perform tasks that would otherwise require human intelligence, such as visual perception, speech recognition, and language translation."

That is a broad definition, but not a bad one. It is a million miles from the "general AI" we might be familiar with from science fiction. Google Translate is not the same as 2001's proactive antagonist HAL 9000 nor even Starfleet's benevolent but entirely passive ship "Computer".

Later in the strategy describes machine learning and natural language processing as AI techniques.

Perhaps, by this point, if you occasionally use tools like Google Translate for work, then I may have made the case that you too use AI professionally.

When I think of AI with my digital marketing daily task list in mind, I think of media targeting such as Google's Smart Bidding algorithms in paid search and on the Google Display Network or of analysis such as Facebook's open source Prophet system for making forecasts or trying to evaluate the impact of a sponsorship deal, a creative refresh or some other quantifiable variable. These are machine learning solutions.

You can find the 11-page write-up of Scotland's Artificial Intelligence Strategy here as a PDF, but I'll give you a summary.

Scotland believes the AI future is coming and wants it to be trustworthy, ethical, and inclusive. The Scottish strategy ensures that businesses and future generations can get the most out of this future and do so ethically.

Scotland will incorporate the United Nations Convention on the Rights of the Child (UNCRC) into domestic law and adapt UNICEF's policy guidance on AI for Children. That means the country will create an enabling environment for child-centred AI, protect data, be transparent and prioritise fairness and non-discrimination.

What does that mean for your website?

I cannot give you legal advice, and frankly, no agency can either. Or should.

GDPR requires some businesses to appoint a data protection officer, and if you have one, they are a good starting point. A businesses' legal counsel or team is another resource.

Personalisation

With that disclaimer out of the way, I want to loop back to the importance of non-discrimination. Any "non" can be hard to prove, so make sure your paper trial is robust. If your site uses dynamic pricing, such as making tickets more expensive as they sell out, then consider documenting the paths and potential outcomes now to show that none of them is discriminatory.

Currently, most personalisation solutions require either the use of cookies or sessions.

Sessions are usually terrible for your SEO. It would be best to avoid them on any site that you want Google to crawl and index easily. Why? Well, sessions are temporary and therefore not suited to the search engine. Sessions URLs are likely unique to an individual and therefore not suited to a general index.

Cookies are unlikely to be considered necessary functionality in this case and therefore require opt-in permission. That brings me to the next takeaway.

Cookies and analytics

Sites in the UK need permission from visitors before they deploy cookies not strictly necessary for functionality, and so you need a cookie permission prompt.

Google already has some solutions. The media giant already uses AI to evaluate whether someone has seen your display ad already, whether Google should show it to them again or not.

The currently proposed replacement to cookie-based analytics is a system known as Federated Learning of Cohorts (FLoC), and, yes, it uses AI too. FLoCs are being tested via the Chrome browser, but not yet in Europe, where there seem to be some privacy laws concerns.

The main difference between the feature-rich Google Analytics Universal and not yet as useful Google Analytics 4 is that it supports FLoCs. Thoughts on GA Universal and GA 4 are a frequent discussion in Leith's digital marketing team, and surely enough for a blog post or three later.

Importantly, there is a lot you can do with Google Analytics right now and hopefully, that's already on your site. It would help if you were building audiences to include or exclude them from your Search or Display targeting. Even paid search (PPC) can make heavy use of audience lists if you tap into Smart Bidding and allow the machine learning to adjust your bids on the likeness of potential visitors to meet your business goals.

For example, let Google's Smart Bidding see which people visit your "Our Services" page and learn whether that's an indication that they're likely to contact you or not. Let the system see which people visit your "Thank you for contacting us" page, and decide whether you want to spend any more money showing them ads.

Ideally, your website will have been designed and coded with analytics capture in mind, and together with Google Tag Manager, you can curate ethically sourced data that adds punch to your media budgets.

Digital Marketing

I've already strayed into digital marketing, haven't I? It's a force of habit. However, getting more value from your Google ads can be as easy as that.

Your social media budgets can benefit in similar ways. Make sure you can deploy Facebook or LinkedIn pixels when you need them, as the law requires so that the social platforms can optimise spend for you too.

Digital marketing best practice such as UTM tagging on social media links so that you can build audience groups from them, creating a virtuous data cycle, should be embraced.

By this point, I think it's clear; getting the most from AI for your website tomorrow means getting your data strategy shipshape today.

Purpose-led

I know "purpose-led brand" is as much a buzzword as "AI technology" is, but that doesn't mean it isn't true. In 2018, Accenture's research showed that 53% of UK consumers prefer to buy goods and services from companies with a purpose similar to theirs.

I know I do. As an anecdotal reference, every person I've shown my Giki Badge app has thought it was a great idea. One quick barcode scan reveals whether a supermarket-bought product has a similar but more ethical alternative.

The Scottish AI strategy is unmistakable with its implication; people are worried about how brands use data and new technologies such as machine learning and natural language processing. I would encourage all brands to have something to say about their use of AI, and I would make sure that was believable, reassuring and appropriate.

Who's doing it right?

There are plenty of examples of Scottish companies doing great things with AI in the Scottish Strategy document. In some ways, the paper is a showcase of Scottish talent in that respect.

There's Alli-chat by Voxsio an app that helps you talk through thoughts and emotions, Kythera's AI for games and the academic success of SICSA, for example. Even while this article was being edited, waiting to go live on the new look Leith site, Michael Rovatsos took to the Scotsman to point out the success Edinburgh University was having in bringing AI talent together.

While the Scottish paper naturally points out Scotland's success, it's fair to say that the UK is doing reasonably well. Tortoise's Global AI Index puts the UK third, ahead of Canada, Israel and Germany, but behind China and the United States.

The UK's success is with what Tortoise defines as the "Operating Environment", and that's precisely what Scotland's AI strategy aims to boost.

The Scottish AI Playbook that will come out of the strategy has three tracks.

Track 1: Establish Collective Leadership.

Track 2: Create the Foundations for success.

Track 3: Build AI Powerhouse.

Track 1 means putting together a team of leaders, and I hope digital marketing is involved in this in some small way.

Track 2 is precisely that "Operating Environment" is so strong in places like Norway, Singapore and South Korea while boosting "Infrastructure" where countries like Pakistan, India, and Indonesia do well.

Track 3 is the promise. And which digital marketer wouldn't want their website to be the front end of a business powerhouse?

A glance around the Scottish technology scene reveals companies of note that didn't get mentioned in the strategy paper. I see GIGGED.ai, for example, the Glasgow-based freelancer platform and Data Innovation.ai based in Edinburgh, using AI to drive a bio-safety tool to help counter COVID-19.

Conclusions

Whether it's sci-fi style AI or the (more likely) business impacts of machine learning algorithms chomping through your data to make decisions and take action in real-time, AI is here.

And people are worried.

Scotland's strategy reflects that concern while embracing the business advantages to be won by doing AI right.

Businesses should design their advertising, marketing and website development with data in mind so that the machine learning tools have access to the information they need. However, this should be done while imagining a mean spirited tabloid is out to find faults with the approach because the brand damage from an algorithm making the wrong call is just as severe as if senior leadership did it. Just ask school kids.

Andrew Girdwood

Andrew Girdwood

Head of Media Technology