“Ultimately, LLM tech doesn’t have to achieve Artificial General Intelligence (AGI) to completely transform the world,” predicts Max Tatton-Brown, Product and Marketing Advisor at human-AI collaboration startup Beyond Work.
“Even if it just automates all the tedious work (high volume, data heavy, repetitive — but ultimately necessary), that will be the biggest evolution for most people’s day-to-day work lives since a PC arrived on their desk.”
Generative AI and the excitement surrounding it was a pervading theme of 2023, with applications in customer service, travel, beauty, healthcare, and potentially much more. But how well is generative AI living up to that excitement? How can businesses who experiment with generative AI balance innovation with risk? And what is the future of human input into AI content?
We posed all these questions and more to a roster of AI experts – here’s what they had to say.
It’s still early days. GenAI is “no substitute for years of experience”
“AI will be a disruptive technological force like we’ve never seen before – and we don’t yet know the full spectrum of capabilities it can offer,” says Fergal Reid, VP of AI at Intercom, an AI customer service solution. “It’s important to strike the right balance between raving about this new technology and remaining judicious about what’s actually happening.”
It’s also important to be aware that “the next 18 months will still be the warm-up phase” for generative AI, in Reid’s estimation. He compares AI’s current status with the early internet: “Looking back at the early days of the internet, people were focused on making simple websites – a long way away from the major companies that ultimately emerged from the Internet.
“Customers will take their time to adopt new technologies, and it’s going to be important for companies to one, meet the market where it’s at and two, invest in building for the future.”
You can ask it for basic tasks, but it’s no substitute for years of experience crafted by a professional.
– Will McMahon, Spark Foundry
Will McMahon, Head of Adtech at media agency Spark Foundry, also appeals for perspective. “Generative AI is still in its early learning years. You can ask it for basic tasks, but it’s no substitute for years of experience crafted by a professional.
“…it’s yet to be seen how good generative AI will be at replicating experts, but it’s initially looking like it will be a great tool to summarise large quantities of data to give much more meaningful insights for business decision makers meaning that we can develop smarter, quicker and more meaningful strategies for success.
“For our teams, that means better insight leading to better decisions – or more creative strategies that lead to better business results for brands.”
Any business risk is human. “AI can only do what you let it”
All businesses want to stay at the cutting edge of technology and innovation, and currently, it looks as though accomplishing this will necessarily involve using generative AI. Indeed, Elisabeth Ling, Product Advisor and Non-Executive Director at eSure Group plc, says, “The largest risk … is inaction, allowing competitors to outpace a company as they harness AI to transform product development, customer interaction, and reshape their business models and financial outcomes.”
However, generative AI’s potential pitfalls are well-known, such as ‘hallucinations’ – in which an AI tool produces convincing-sounding errors – and copyright implications that are still being litigated in the courts. How can businesses balance the desire to innovate with staying clear of these risks?
“Ultimately [generative AI] can only do what you let it,” says Max Tatton-Brown. “That should be a design and implementation challenge handled by whoever is helping you implement it or is building the software. There’s no need to fear this tech going ‘rogue’ or accessing anything you haven’t let it.
“As usual, the threat is actual human beings. Making sure people aren’t doing things in an unauthorised way, or aren’t doing things off your radar. But that’s something all organisations should be paying attention to with their use of tech anyway.”
Ultimately [generative AI] can only do what you let it … As usual, the threat is actual human beings.
– Max Tatton-Brown, Beyond Work
Spark Foundry’s Will McMahon agrees: “It makes mistakes and can hallucinate, so all work should be thoroughly checked and cross-referenced. … [W]e’ve put rigorous human checks in place for any generative AI solutions to ensure that we’re not the victim of hallucinations or misinformation.
“That said, the time it takes to check a mundane task is much shorter than doing the task itself, meaning that our teams can spend more time doing the harder jobs that AI isn’t equipped to do, such as planning and buying best-in-class media campaigns.
“Second, there’s a real temptation for employees to put sensitive information into generative AI agents, which could breach data privacy guidelines or laws. For that reason, all businesses should be extremely clear about what can and cannot be shared with generative AI.”
Download now: The Future of Marketing: The AI Opportunity : findings from 2023’s Future of Marketing survey.
Human involvement with generative AI is “absolutely fundamental”
As generative AI becomes more sophisticated, what will be the future of human interaction with AI output? Will there continue to be human involvement at all? Tatton-Brown thinks so:
“It’s absolutely fundamental — there’s no point in doing this, other than to free humans up to spend more time on the work that actually matters and turns the dial.
“I think we underestimate the cost of having to split our attention and waste human-level thinking on tedious tasks that should be automated. Expenses, invoicing, onboarding team members, handling requests, trawling through data to find insights.”
McMahon similarly envisions AI bringing out the best in human potential: “At the moment, people are still getting to grips with how generative AI allows them to do some of the more tedious parts of their jobs a bit quicker, but in 2024 I think we’ll start to cement the best use cases. This will lead to more people using generative AI across the business, allowing them to apply their expertise more effectively for brands.
“But it’ll also mean that we’re testing, innovating and, crucially, sharing more and more uses, which allows us to expand our roles and provide better, more impactful work for our brands.”
To facilitate this, Kate Cox, Chief Marketing Officer at BrightBid, emphasises the importance of getting your data in order. “It is incredibly important to set up marketing processes for the AI future. …to use it effectively, data needs to be structured in an AI-friendly way and with an underlying logic.
“Attention needs to be paid to first-party and customer data structures as the foundation of your data strategy; the linking of data sources to uncover insights; and setting up data for effective testing – test/control or A/B tests.”
Upskilling for the AI era
As businesses experiment with the possibilities of generative AI and the technology develops rapidly, how can teams make sure that their staff have the right skills?
“The AI universe is moving so fast that only those really immersed in the industry are truly au fait with developments,” says Cox. “Everyone else needs training on specific systems, and that training needs to be updated every time the stack is altered.”
However, Cox has some practical advice to offer for marketers, who have the advantage of mostly using low-cost AI tools or tools built into existing tech stacks such as Salesforce and Microsoft (as opposed to more elaborate AI setups for back-end operations or business processes). She recommends that marketers “dive in from a personal point of view and start using the tools and experiment with how these could be applied for work use.”
This approach means defining the “jobs to be done”, “testing and learning by channel” and noting “productivity and quality improvements”, but also managing expectations as you go.
What makes this AI ‘moment’ unique is its accessibility
“Artificial Intelligence (AI) in general, or rather the application of mathematical models and algorithms in business applications, has always had tremendous potential,” says eSure Group plc’s Elisabeth Ling.
‘AI’ in some form has been around – and used in marketing – in some form for years, but Ling argues that what makes this iteration of AI so exciting is its simplicity and accessibility. “The ‘hype’ [around generative AI] stems from ChatGPT’s revolutionary approach both from a model AI standpoint, but also from the simplicity of its interface and unprecedented consumer success. As millions embraced it so rapidly, it is only normal that this created buzz,” she says.
Tatton-Brown agrees: “…this is the first technology where you don’t have to learn its interface to use it — because it knows how to use ours (language). That means no relying on IT teams to set up automation for you or get the value for you. That’s massive.”
Econsultancy offers a short course in AI for marketing