Ask not what generative AI will do for you, but what together we can do for the freedom of AI
Unless you’ve been hiding under a Wi-fi resistant rock the last eight months, you’ve probably heard generative AI is the hottest tech trend. You’re bombarded by rose-tinted utopias about how it’s coding in half the time, or using first-party data for personalized content creation.
In camp apocalypse, AI experts talk about dehumanization, job decimation — and, inevitably, human extinction. Frankenstein’s remorse now their creation is out living its best life? Or the logical conclusion after deep thinking about the impact of any technology?
For sure, there are some bad car crashes on the generative AI road, like a fake AI interview with Michael Schumacher that didn’t thrill the relatives of the, very much alive, racing driver.
What of its promised utopia — newer, faster, better?
Recently I talked at Unlimited B2B’s Humans Vs Robots: The Future of AI in B2B Content. The show of hands in the room indicated most marketers are already on the generative AI road. But are we heading in the right direction? At the right speed?
Right now, generative AI is less fast-track, more off-road terrain. Be prepared for accidents on route, and a shifting ETA or even destination.
Event host, tech journalist Martin Veitch, says generative AI is still in its infancy. Dr. Christina de Balanzo, neuroscientist from Human Understanding Lab, advocate for the human in our debate, thinks infant AI can behave like a toddler. We need to be careful about what control we give it. As humans, we’re hard-wired to resist change. Is this a trend to big hug or treat with kid gloves?
We need to go beyond amoring AI as the shiny new thing and tune into what it can do for us. To paraphrase (bastardize) John F. Kennedy:
My fellow marketers of the world: ask not what generative AI will do for you, but what together we can do for the freedom of AI.
Generating order from chaos
Generative AI works with unstructured data, “66 billion data points, retrieved in seconds” as UNLIMITED’s technical director Tom Wilks, our panel’s advocate for the machines, explains. All that big data overwhelms us delicate humans, but AI can shape it into a format we can use. Like clay dug deep from the earth processed into a substance for the potter’s wheel.
Build a chatbot to serve customers the right answer to frequently asked questions? Check.
Code your app with a new interface to help people use your new product? Check.
Help a clinician identify an obscure disease from a patient’s symptom? Check.
Assist lawyers to sift past case law files to defend clients for lower fees? Check.
Use cases where AI structures existing information, or adapts things in a new context, puts the intelligence into artificial intelligence. Now what about the generative bit? What can we trust this novice, slightly wobbly, tech to make for our business?
There’s brand and user safety to consider. AI is already spreading disinformation like fake reviews, deepfakes, and ‘creators’ using generative AI include cybercriminals.
Top of mind for marketers, AI copyright is a hot potato. Comedian Sarah Silverman sues OpenAI for using her book to train AI using illicit ‘shadow libraries’. A radio host sues ChatGPT for defamation. Getty ain’t too happy with Stable AI for ‘borrowing’ 12 million copyright images.
For business, not breaking copyright law can be pretty important. But generating IP may be super important too. According to the US Copyright Office, you can’t copyright AI-generated images, text or video. Sands may shift. A similar argument was made in 1835’s invention of the camera. The machine captures reality, the photographer merely its processor. Until the late 1990s, the Musician’s Union in the UK banned players of synthesizers, samplers and DJs as people not considered ‘real’ musicians.
Thankfully for Man Ray, Cindy Sherman and Ian Rankin — and commercially for Getty Images and Instagram — photography as a creative discipline stuck. And with new creative forms comes the money.
I like big bucks
McKinsey predicts generative AI will be worth up to $4.4 trillion. But this time last year McKinsey ‘I’ll have what they’re smoking’ pipedreamed the metaverse’s value at $5 trillion, a stat neatly pulled apart by James Whatley in my podcast, Insight Story.
Money is power. The big bucks headline could be a big opportunity. CFOs salivate at automation making business ruthlessly efficient. Unlike the metaverse’s alternative reality, AI feels like a reality-centred deal for belt-strapped times. But to get great efficiency takes great skills and investment.
As we discussed in Insight Story’s podcast about AI, only 3 in 10 AI projects reviewed by MIT in 2019 saved any money. That may be because projects were lurking in the R&D lab. Recent outcomes are more promising: Descript’s July 2023 report surveying podcasters showed AI adopters reported higher incomes. Early green shoots, though I’m a tad cynical of a survey by an AI tech firm. The report’s got good advice about using AI tools to provoke new or unexpected ideas, and shortcut repetitive marketing tasks. This could be the road to our utopia.
For marketers, we’re expected to do more — more channels, volume, testing — often with the same resources. Generative AI could be our savior. But only if we can use it to augment our creative brains. This can’t just be a productivity play. If there’s a CMO who gets fired up about super efficiency and downsizing creative roles, please stand up. You know where the door is.
Without due consideration, it’s not just brainless tasks that could disappear. Skilled jobs that take expensive training to understand all those data points like law, finance and engineering are most at risk, according to the OECD. Could mass automation lead to even more menial work feeding the machine?
As a brand content expert, right now, I have equal intrigue and equal fear and loathing. My curious self knows AI could be great, and I do want to get ahead, but it isn’t so great yet. And my perfectionist self can’t quite reconcile that.
My dabbling with Chat GPT found super obvious, sometimes wrong, answers to topics I know about. Give it a creative research brief and the deck may look and smell like what you pay your creative agency for, but the ideas are yawn-inducing. Billions of data points can only offer up what’s gone before. And good luck fact-checking copy created from tools with unidentifiable data sources.
What tasks are generative AI good enough for?
If you’re a risk-averse (read: copyright upholding) business, what’s your options? There are tools you probably don’t think of as generative AI that can optimize workflow. Like spelling and grammar assistants, machine translation and formatting copy. These tools have advanced so much in recent years when offshoring audio transcription was the only option as automation produced garbled nonsense. The gap between good enough and not good enough is vast. An 80% correct translation is 100% useless. A 95% correct one is 95% useful. A knowledgeable human can fill the gaps and adapt for their audience.
Tell me ideas that have gone before, so I can take those out. And if you work with an agency that generates obvious ideas, ask them to do better.
Give me a structure and outline for a thought piece. For non-writers with ideas, it’s a brilliant boost.
If you’re a new or scaling business that needs lots of ‘content stuff,’ AI is a budget motor to get moving quickly, albeit not gracefully.
Generative AI does not sleep. But it also doesn’t care whether it’s changing hearts and minds. It may be great at writing linkbait headlines or a call-to-action button that demands to be clicked. But is the story compelling? Does it improve the perception of your brand or leave readers feeling ‘meh’? Will it win an Effie or Cannes Lion award? Probably not. Does it have to? Only you can say. Good enough may be good enough.
We dabbled using generative AI images for Secure Futures magazine — fittingly, for stories about AI and fake news. The results were so-so. Telling an immediate visual story about a complex emerging technology, as it turns out, is complex work. I’d rather pay real writers and artists to help detangle these topics. They may choose to use generative AI prompts to get started or support their work. We aren’t reinventing clay here, but perhaps igniting the potter’s wheel to form new shapes.
I may end up the last creative marketing director standing lonely on the hill of human creativity. But for enterprises that can afford it, I think it’s right to hire the best people to do their best work.
Go fast. But don’t break things.
The journey begins. But it doesn’t mean you need to go fast. Just go right. Let’s not make this another ‘disruptive’ tech that moves fast and breaks what came before into shards.
One event attendee’s business embraced it wholesale, with business, marketing and tech teams working together to establish the guardrails. This makes me very happy. To make new tech work, we need open communication, plus some boring stuff like policies and processes defining how it supports the business. Ethics and copyright should be on the agenda as frequently as licenses and training. Let’s not hire ‘prompt engineers,’ but rather use experts to train your teams to use tools the right way.
Another event attendee uses AI to generate snippets for SEO to land traffic on a web page, thankfully, written by a real writer. The machine feeds the machine, to serve the human. Like a conveyor belt in a robot restaurant. Maybe that’s smart marketing.
My biggest fear for content marketing? We’re already drowning in a sea of sameness. Do we need more dross, now amplified by machine? How can we give creatives the freedom to unpeel and elevate above it? Automation could help structure that strict 3,000-word whitepaper or social adverts with fiddly character counts. If you can’t find or afford creative greatness, it could be good enough, or your best available option.
My second fear is in the cliché trope, ‘robots are taking our jobs’. When creative students graduate from AI-led education into work, will they get mentored by experienced pros to know how to craft greatness from the noise? I’m sure the old guard Exec Creative Directors already say digital killed the ad star. But without new inspiration, the creative pot quickly becomes shallow.
Like a ChatGPT response, thinking about this is leading to more questions than reliable answers. But like pottery, the creative discipline I’ve never tried but oft dream about before as it’s the least impacted by technology, I know it’s a fever dream. I later ran digital marketing training for craft professionals. Because their world had finally been digital disrupted.
There is hope for some of us humans: as The Ivy’s bar manager at our event told me, he’s not worried about losing his job. A robot bartender would not work. It misses the human cues and our needs. (Edit 25 July: Eeek. Looks like in some contexts the robot bartender is here). A robot content director, though — could go either way.
The future is here. We can’t be industrial age Luddites breaking the machines. But we can’t become slaves to the digital dictator.
Go fast. Go lightly. But please, don’t break things.
Are you using generative AI for content? How’s it working for you? What works and what’s making you gaga? Let me know in the comments. We’re producing a new series of Insight Story podcast with episodes about Generative AI and AI ethics. Get in touch with suggestions for guests you’d like to hear.
