If you look at industry surveys on the adoption of generative AI, you can see that many marketers are slowly venturing into the topic – or at least dipping their toe into the deep end.
However, the easy access to these productivity tools means that their adoption can sometimes be ad hoc and fragmented.
For example, Adobe’s 2024 Trends Survey found that only a quarter of client-side executives had already undertaken genAI skills development programs. And yet, new data from Marketing Week shows that nearly half of marketers use AI for market research and more than two in five for audience segmentation and creative testing.
I’m sure we’ve all talked to people in marketing and other industries who have struggled with using new tools to create images and text and advised others to do the same.
Is it a problem to be open-minded and keep up to date with new technologies? Isn’t that what every good marketer should do?
Will GenAI save time or will it overwhelm marketers on a large scale?
Is it beneficial for the entire organization?
There is a risk that ad hoc experiments may only deliver small local benefits while missing the opportunity to deliver larger benefits for the entire organization.
I spoke to an experienced data and analytics leader earlier this year who used the analogy of the early days of CRM systems in B2B companies and the difficulty of ensuring adoption among sales reps.
“For more than ten years (companies said) they had managed their contacts and key sales processes in CRM. They never did it properly. And then suddenly they realized they can manage their contacts on their smartphones too. They can find customers, use WhatsApp and start their own little campaign within minutes. They create contacts, customer groups… So basically the company has no access to this customer data. It’s on people’s devices. It’s fragmented. It’s lost,” they told me.
“If they had developed a good strategy for a CRM system and found ways to capture that data, they would have a wealth of customer data. They would have a unified record. Their customers would be happy, the sales reps would have the productivity they need, and the asset would be managed as a company asset.
“I’ve seen many companies miss this opportunity. The same thing could happen in AI.”
The risks of free agency
The potential disadvantages of allowing a salesperson, marketing, or customer service representative to incorporate Large Language Models (LLMs) into their workflow are well documented.
CEOs may view GenAI as a threat to compliance and competitiveness, often focusing on issues such as hallucinations and privacy violations.
Providers are also aware of the challenge of aligning the use of GenAI with corporate strategy and strengthening institutional knowledge.
Salesforce, for example, talks about “dynamic grounding” on marketing pages for its Einstein Trust Layer, which is described as adding “domain-specific knowledge and customer information” to prompts to generate more accurate responses, for example by using “CRM data, knowledge articles, service chats, and more” to “reduce the likelihood of hallucinations.” There are also details of “data masking” and “zero retention architecture,” which shields data from external LLMs.
In the case of ad hoc use of LLMs, one could argue that productivity tools are by their nature scattered and that employees must already take personal responsibility for their use of everything from Google search and social media to enterprise software.
There is a risk that ad hoc experiments may only deliver small local benefits while missing the opportunity to deliver larger benefits for the entire organization.
This is undoubtedly true, and the first step in training is to ensure employees know what constitutes responsible use of new tools. But the broader point remains: is there a need for AI to solve a business problem or not? If so, then that solution needs to be aligned with the company’s strategy.
Cassie Kozyrkov, CEO of Data Scientific and former senior decision researcher at Google, wrote in a post on LinkedIn last month that AI “should be what you try when traditional programming has failed. When you need to automate something but are unable to do so with your existing repertoire. When the need is so great that you are willing to increase complexity and the reduction in control that comes with it.”
When asked whether GenAI tools increase the efficiency of all organizations, Kozyrkov responded in the comments below her post: “Poor management has been shown to reduce efficiency.”
Expectations for GenAI are enormous: Two-thirds of executives in Adobe’s recent survey expressed optimism that the technology will lead to business transformation in analytics, content, customer service and sales. But without strategic oversight and if people simply follow their own initiatives, those expectations are unlikely to be met.
Ben Davis is Insights Editor at Econsultancy, which delivers e-learning, live learning online workshops and skills mapping in digital, marketing and e-commerce.
