AI and Creative Technologies (Communications and Digital Committee Report)

Earl of Erroll Excerpts

Friday 13th June 2025 Lords Chamber

[The Earl of Erroll (CB)]

My Lords, paragraph 154 made me think about what on earth I could usefully add to this comprehensive and very pertinent report. We were consistently reminded that the creative industries sector is made up almost entirely of SMEs, of which a high proportion are micro-businesses. I have spent most of my life working in SMEs and building them up. On the technology side, I used to write software and things like that. I was always involved in computing IT, trying to produce solutions for business, looking at practical applications of things and trying to develop solutions quite quickly. It is this application of technology that most interests me.

I agree with the noble Lord, Lord Holmes, that the most useful applications of AI will not be for writing nice shareholder reports, business presentations or applications for jobs. It will be for managing real-world, complex situations and applying huge amounts of data and information to get out what needs to be done and what reactions. You could use this for managing difficult things. It can also be used for analysing huge volumes of research if you do not have time to read it all and trying to make the links between the different bits of it. Here, the LLM stuff will help considerably.

I did a lot of work looking at IoT—the internet of things—a few years back. In fact, we produced a British Standard on data interchange for it, because the use from it came from when you combined information from different sources and sensors around the place to produce something useful for people. That has grown into an interest in digital twins, where you can mirror a real-world thing out there in a computer system and then use it to analyse what would happen if this happened, or what is happening right now and how you might respond to it.

I am about to have an involvement with the Connected Places Catapult, which is mentioned in the report. I think it is very important that we continue to support these places, because these are sources of innovation where lots of other people are putting in lots of ideas and things, and out of those have come some very useful things. The great thing about them is they have been consistent—they are still there.

One of the big problems that I have noticed—I will jump to it—is that there is always a plethora of grants, but they are there for a year or two and then disappear. There is no consistency, so you cannot plan for it, and things collapse as a result. The creative industries clusters programme is a typical example of that: just when it is working, I think the report said that it is being knocked on the head. What is the point of that? At government level, we always seem to ditch things just when they are showing success, and that needs to be changed. Sometimes, people have unrealistic expectations of what will come and how quickly. That is one of the big problems, because Governments, Civil Service priorities and departments change.

I have also got very interested in causal AI, because I realised we have large language models just chasing down word chains and putting them together, with no idea what they are looking at. There is no actual innate intelligence in artificial intelligence: it is just a very large neural network of transistors, and it does not work the way our brains do. It does not have any sense of empathy or history. It has no background, unless you have trained it, but there comes a limit to how much we are going to train it and whose learning we are going to train it on. I know that most of us think we are always right, and that is always the trouble, because we do not agree with how someone else thinks. That is part of being human. I am going slightly off on a tangent, but there are some limitations in AI there, and people need to be aware of them.

The point about causal AI is that a lot of this information coming from the big ones—Copilot, Gemini, ChatGPT and all that—may be drawn from anywhere. You just do not know what they have looked at. There are lots of things that are false, and that self-reinforce, out there on the internet, so it might come up with things in reports—particularly if you are relying on it for something—that would completely mislead. You need to be able to distinguish that. I have been involved with a company, Kaimai, as well as FIDO, which have been used on curated databases to try to extract data. We need to grow that using causal AI, because things change as well. You may have a good database, but stuff that went into it 10 or 15 years ago may be inaccurate by now—the world has changed. There are a lot of issues in there that people are not thinking of when it comes to these things.

On encouraging people to stay here and grow their businesses, why do we drive all our very successful people offshore with very high tax rates? At the moment, Dubai is doing very good business, and apparently quite a lot have gone to Italy since the great non-dom thing was re-echoed. People want to spend their money when they make it, and that keeps the economy going, because it keeps all sorts of other things going, such as expensive restaurants, people who make expensive goods and people who cater for all sorts of things like that. It also makes the place more interesting to live in: people want to be in the UK, or around London, because it is full of entertainment. There are all sorts of side benefits to keeping your people who are successful and not driving them away. This is quite apart from the fact that I get furious when I see our great successes swallowed up by large American corporations and watching them go offshore.

Sorry about this, but, typically, my device has gone to sleep—like me.

Another thing I think about is suggesting that we can get the pension funds to invest a certain amount of money and push them into doing it. Yes, that is a source of funds, but are the pension funds the best people to decide what is a good or bad investment? Their job is to try to make sure that—if pensions are not overtaxed yet again—you have some money to retire on, and to make sure that the money is there for you when you retire. I am not sure they are the right people to judge what to put money into. I am sure that someone will think of a good way around this, but we have to be very careful about horses for courses.

The noble Lord, Lord Evans, made a very good point about regulation stifling investment and research. That can happen a lot. Funnily enough, another thing that can stifle some of this co-operative approach, which we hit a bit back, is that some academics in universities really do not feel that commercial applications of their knowledge and learning is the right thing for academics to be involved in, and they would not co-operate because there might be a commercial outcome from it. I think that attitude is changing—I have not had anything to with universities—but I bet it is still alive and well in some places. Those sorts of attitudes need to be overcome.

Here is another thing we hit: although we were being funded by the Welsh Government for an innovative programme, we needed a good native foreign language speaker. It was in the early days, before these were called AI and LLMs. It was to do with formatted reporting and stuff. Could we get the work permit for them? Well, we eventually got permission to employ one person from abroad—they were an ex-student from the university, but we needed to get the ongoing work permit—but were not given the permission to apply for a visa. That was going to be another application, and more money. These sorts of bureaucratic things kill SMEs. We just do not have the time, energy or knowledge to get around them. We need to start thinking about that.

The noble Baroness, Lady Kidron, made a comment that data sharing is essential to get the best use out of it all, but there are huge dangers, as I have just been saying, about where that data has come from and what happens to it. If it is being used by some abroad to do all sorts of things, it can be hugely dangerous, even to our national security and other things like that. You never know—even the best people have something to hide and, if that comes up, you are opening people up to maybe a bit of blackmail or pressure. That is why I have been very cautious about government data sharing in the past. It is difficult: it is impossible to pseudonymise properly.

Anyway, with that, I think it is a brilliant report.