Thought Leadership

How should we think about AI? – The Sorcerer’s Apprentice

You might recognise the story of the sorcerer’s apprentice from Disney’s Fantasia. An apprentice with laborious chores to perform enchants a broom to help him, only for the magic to become out of control, the ensuing chaos only ending when the sorcerer returns, cautioning only an expert should invoke powerful spirits.

After a period of exponential development, artificial intelligence is consolidating. But it has already hit the mainstream. The uncanny, fuzzy, six-fingered mannequins are being ironed out, and generative AI is being used for everything from research papers to eBay listings and Instagram reels.

Over the past year there’s been think pieces, hit pieces, opinion pieces, thought leadership, soapboxing, fear-mongering and grandstanding. As AI begins to integrate into our daily lives, it’s time we take a breath. Take stock of the more rational, unbiased voices. Look at how we’re using AI now and consider where we’re taking it.

The Shades of Intelligence report by ItsNicethat in late 2023 found that 83% of creatives had already adopted AI into their working practices. It’s still one of the most insightful series published on generative AI. We’re beyond the point of questioning whether it’s legitimate. In our workplaces and digital lives, AI is here to stay. It’s the early days of the adoption of this new technology. And worth taking a wider lens, medium term view.

While avoiding the phoney comforts of AI scepticism, generative AI is still error-prone and producing a lot of mediocre output. “AI slop was shortlisted for the 2024 Oxford word of the year, referring to the vacuous phrasing popping up all over the internet that has become a poker tell for AI copywriting. At times the output of what has been touted as the great technological leap of our time is overconfident and misinformed; prone to hallucinations and prioritising algorithm-friendly sludge over results of value.

Those like Templo, working in arenas where truth and accuracy are as important as efficiency, are limiting generative AI to specific tasks and making more use of its data mining capabilities. Without human oversight, AI compounds algorithmic biases of race and gender and spits out insensitive and offensive content. “There was the case of black people being classified as gorillas; the computer system that rejected an Asian man’s passport photo because it read his eyes as being closed.” (Artsy)

Sectorlight has two offices, covering two continents, working across time zones and cultures. Having local experts that understand the regions where our projects launch is vital; overlooking cultural nuances has a massive impact. That contextual sensitivity and quality control is something we bring to all our creative work whatever technology we’re using.”

Danny Brooks – Digital Creative Director

Then there’s more intentional misinformation. Viral examples like the Pope clad in an ostentatious white puffer jacket have been upstaged by in-person events contrived by phoney AI promises – like the memorably unofficial Willy’s Chocolate Experience. The spectre of dark web ‘deepfakes’ haunt the news cycle and herald an era of misrepresentation. Globally, lawmakers are rushing to catchup. The legal basis of copyright has been turned on its head. Midjourney’s interface itself admits we’re in the ‘wild west’ era of AI. Given how long it took the law to catch up with the invention of the internet, it may be some time before we can be comfortable AI is under control.

Listening to leading scientists on the long-term, 30 year view, things get a bit existential. The paperclip theory strikingly illustrates what’s known as the “alignment problem”—the discrepancy between the ways machines complete tasks and what’s actually good for the humans who assigned those tasks in the first place. We’re not at the stage of superintelligence just yet. If AI is an existential threat on par with climate change then we need at least the level of corporate and governmental responsibility environmental initiatives have grudgingly been afforded.

This ItsNiceThat article points out, “unless you’re off the grid, you’re already using AI whether or not you see it.” Though the uncanny images and cumbersome LLM of generative AI have captured the public’s imagination, analytical artificial intelligence is naturally suited to processing data, making systems more efficient and saving us time. In real estate, JLL recognise analytical AI as ‘table stakes’, increasing efficiency with document processing, investment modelling and risk management as well as gen-AI chatbots and virtual tours. As John Naughton puts it “Information is the fuel on which corporations run, and they will adopt any technology that provides a more intelligent way of handling it”.

The automation paradox states that with this paradigm shift human involvement becomes more critical, not less. ‘The more efficient the automated system, the more crucial the human contribution of the operators”. Being able to extract useful content from the AI slop, to sort the wheat from the chaff, becomes integral to the machine. In the words of ItsNiceThat’s editor “AI can’t give you good taste”. It requires skilled and knowledgeable operators that know when and when not to use it.

It’s worth remembering “that AI isn’t a monolithic entity, it is really a collection of specialised algorithms and models designed for specific tasks.” By controlling their datasets and fine tuning their architecture and parameters, individuals and companies can gain access to unique informative insights and IP control. Graphic designers have already began coding their own unique AI tools, while exactly.ai “empowers creators to train ai models using their own artwork, retain their copyright, and earn revenue for their work”.

Forbes predicts “the creators who thrive will be those who can effectively harness AI’s efficiency while maintaining the human elements that make content truly engaging.” This will continue the trend towards smaller studios existing tech has induced. More agile teams, specialised to markets, capable of selectively using AI as part of a human feedback loop.

Perhaps the most appropriate fictional illustration of AI is found in two iconic sci-fi droids. C3-PO and R2 D2 are helpful, calculate odds and are prone to faux pas and blunders. They have hyper specific capabilities and are vitally important copilots and assistants. But they also need to be minded and are not always deferred to.

Jacob James, our Design Director, explains the role AI plays in our agency: “How we use AI in the studio and to what extent depends on the program and the task at hand. Plugins like Adobe’s Generative Fill have become the norm for speeding up art working. At the ideation phase it can produce visuals quickly, and if you spend time on prompting, can throw up some unexpected and unique ideas and combinations we wouldn’t necessarily arrived at.

Largely we use it as a jumping off point, we’re careful not to give it too much responsibility. It’s sped up a lot of workflows, but it does have to be managed carefully. Most software can still be hit or miss so it needs studio guidelines and oversight checks.

With moving image its inspiringly useful for storyboarding and testing quickly and generating professional voiceovers. Time and cost-heavy processes like photoshoots and CGI rendering can be prepared for with much greater specificity. In place and space being able to quickly animate static CGIs to pan around an interior is a massive asset. We’ll currently use traditional methods for the finished article, but as the technology improves it’ll be able to move beyond the cost-saving prototype stage.”

Some creatives have referred to AI as simply another tool in their arsenal, but this understates its scope. AI’s potential is vast and evolving. Who could have predicted in 1999 how we use the internet today, the transformative shifts in our personal and professional lives? AI promises to be an empowering assistant, across the challenges, vocations and moments that make up our lives. A co-pilot that helps us focus on what’s important to us.

The debate on what AI can and can’t do can quickly get stuck in current hiccups or speculation. Creatives are at the forefront of how it’s being used here and now. Poking at it, playing with it. With the curiosity to keep probing and testing its outer reaches, with the knowledge and expertise to see where it’s lacking. Understanding its limitations and making use of its strengths. Forming a powerful partnership between creative human and AI co-pilot.

Thoughts