In the second part of his series Douglas Perman examines the risks AI may pose to a trio of issues: copyright, the environment, and employment. Other issues such as privacy feature in Part 3.
I’ve written in Sceptical Scot before about the significant challenges faced by the creative industries in Scotland since the pandemic. The economic challenge has not gone away. Costs are rising but funding (be it sponsorship, commissions or investment) is scant and confidence low. We can use AI to leverage opportunities, but we need to confront perceived threats first.
They are daunting: copyright infringement, environmental impact, job displacement, privacy and security concerns, bias and ethics and (ahem) annihilation when robot overlords take over. All but the last one have validity. Let’s address the first three.
Inner Ear’s live coverage of Perth Theatre’s 125th anniversary street party, the kind of creativity AI cannot do: photo Dougal Perman
Copyright
Compromising copyright should bother anyone who creates or works with intellectual property. In the creative industries, that’s everyone. LLMs are trained on the entirety of readily available human knowledge. It’s the readily available bit that’s causing friction because it includes proprietary copyrighted works by living artists and creatives in all disciplines.
LLMs are excellent at roleplaying. They function best when you ask them (in your prompts) to take on a specific role or act in a certain way, which can include emulating styles. So you can get the LLM to create new work in the style of writers, visual artists, composers, performers, coders, designers, craft makers and so on. This can lead us down a path where the copyright is far from clear.
However, there is a distinct difference between inspiration and imitation. We’re all influenced by those we admire. Influence has been part of the creative process since early humans sang around campfires and painted on cave walls. But when does imitation become theft, and is theft always such a bad thing anyway?
“Good composers borrow; great ones steal”, said Igor Stravinsky (probably). I love that quote. Imitation is fine in the context of sampling, collaging and post-modern satire. But it’s definitely not OK when it comes to passing off, fraud and infringement. Copyright is a complicated, thorny topic worthy of a dissertation — and that’s without considering how LLMs fit into the equation.
Environment
“AI is terrible for the environment,” say some, “We won’t be using AI due to its environmental impact” say others, “every conversation with ChatGPT uses a gallon of drinking water” (and so on).
High-powered processors used to compute AI algorithms consume large amounts of energy, run hot and need cooling. But so do many tech tasks. Writing in the FT, (How AI might save more energy than it soaks up), Camilla Palladino suggests: “AI’s reach is likely to be larger than its energy footprint. The technology is expected to improve the efficiency of almost everything we do.”
AI does use electricity and water, especially concentrated in data centres and in model training. However, today’s best estimates for a single ChatGPT-style prompt response are small: roughly 0.3 Wh of electricity and 0.32 ml of water per query. Scale is the issue, not the per-use impact.
Palladino references the Energy and AI report by the International Energy Agency. The IEA acknowledges AI’s significant energy footprint but projects that its ability to create system-wide efficiencies in power grids, transport and industry could ultimately save far more energy than it consumes. The IEA says its “major new IEA report brings groundbreaking data and analysis to one of the most pressing and least understood energy issues today”. It’s worth taking the time to understand the issues. In the IEA’s data, by 2030, industry, electric vehicles, appliances and space cooling (air conditioning) will all consume significantly more energy than data centres. And data centres don’t just compute AI. Think about that the next time you’re binge watching Squid Game, streaming Chappell Roan, scrolling TikTok, playing Minecraft and reading Sceptical Scot.
If you’re sceptical of AI on environmental grounds, you’re right to ask tough questions. But abstention won’t make any tangible difference. Instead, let’s find solutions to these problems. They can present opportunities. The evidence suggests measured use plus better procurement beats abstention. We should push vendors for site-specific data, choose which models we use for certain tasks, point AI at problems where its savings are tangible and lobby for industry regulation and best practice.
Credible energy bodies argue that AI can deliver system-wide efficiency savings across grids, buildings, transport and industry greater than the energy it consumes. But those savings are not automatic and depend on policy, design and behaviour.
Employment
Numerous predictions warn that AI will cause mass job losses. AI won’t cause redundancies; HR decisions will. Automation shifts tasks first, jobs second. The outcomes depend on management choices.
In 2022 Klarna cut around 700 full time equivalent roles during a downturn. In 2024 it launched an AI assistant that the company says now does the work of 700 agents with comparable customer satisfaction and faster resolution. Results are contested in places. There are reports of selective rehiring while Klarna is running a hybrid human+AI model. Elsewhere, the Commonwealth Bank of Australia walked back AI-linked cuts after union pressure and operational issues; rushed replacement can backfire.
I often argue that replacing people with autonomous AI agents is a false economy. Yes, you can save money, but if you empower your existing people through AI training, you get the best of both worlds and make more money.
Unfortunately, low-level white collar jobs (in legal, finance, software development, media, etc.) will be hit hard. If paralegals, bookkeepers, junior developers and reporters are replaced by AI tools, it creates a present employment problem and future industry development crisis. Senior, experienced professionals may be safe now, but what happens when they retire?
The creative industries have more resilience but I worry about freelancers. Video and design work in corporate communications and mainstream advertising that was previously subcontracted to freelance creatives and boutique studios is increasingly being done by agencies and inhouse brands teams.
As a creative producer, my team and I are able to do more, better, faster; we’re expanding and experimenting with techniques we weren’t previously able to do. Ideas I’ve never had the time, headspace or resources to realise are now coming to life. We’re directing AI to mine data, collaborating with it through vibe editing and vibe coding (i.e. conversing with an LLM that then carries out instructions or writes and amends code to create new media productions or apps) and accelerating our on-the-job problem solving. I encourage everyone I work with to take advantage of AI technologies to empower their creativity (even, and perhaps especially, when the task you enlist the AI to do isn’t itself a creative one).
What jobs are safe? Former chief business officer for Google X Mo Gawdat believes most jobs are vulnerable. He suggests we should be plumbers, for the next five years at least. But eventually even skilled manual work could be replaced by AI robots. Although, I wonder if fixing the plumbing will really be high on the priority list for the robotics industry. Pioneering natural language processing strategist Dr Joan Palmiter Bajorek agrees: “of all the different robots I’ve seen, the plumber robot, the electrician robot, is gonna be a very expensive robot to make.”
While I find generative AI tools tremendously useful, I have rarely read, seen or heard any AI-generated art that has made me feel anything. However, at a Scottish Music Industry Association round table discussion event recently, a singer-songwriter I respect greatly said he created songs using Suno.ai that genuinely moved him.
But I’m not too worried about the long-term impact of AI on the creative industries. Generative AI does threaten low-level creative production, but original art, storytelling and anything with integrity, authenticity, provenance and liveness can only ever be enhanced by technology, never replaced.
I’m with Mo Gawdat, speaking on Diary of a CEO: “I think we should all be musicians. We should all be authors. We should all be artists. We should all be entertainers. We should all be comedians… These are roles that will remain.”
My son asked me if AI would take my job. As a creative producer, director and strategist I’m not concerned personally, mostly because I have a good understanding of how to use AI to empower my work. But I do think we all have a responsibility to reconsider our roles and the reasons we do them. If low level skilled work is replaced then we need to work out, individually and collectively, how to use that surplus human intelligence and ability to do new and better things. However, it would be a grave mistake to ignore the possibility of impending large scale unemployment without reskilling and reassessing our human resources.
Ultimately, whether “AI takes your job” isn’t up to AI. It’s just software. We need to take responsibility for the actions and decisions made in the name of AI.
Part 3 looks at responsibility, how and why we need to take control of our use of AI. Click HERE
Featured images by the author from Inner Ear’s live coverage of Perth Theatre’s 125th anniversary street party – a spectacular show !
Leave a Reply