• by Laurie Sullivan , Staff Writer, Yesterday

Marketing technology leaders have hoped artificial intelligence (AI) agents will alleviate some constraints, with 94% of high performers thinking the agentic technology will deliver significant operational efficiencies, and 98% think it will deliver significant business performance results. 

The reality remains that many organizations still struggle to see significant return on investment (ROI) from basic generative AI applications, according to Gartner.

Chris Willis, chief design officer and futurist at Domo, pointed to some interesting data. AI's share of marketing activities has nearly doubled in two years, from 13.1% to 24.2%, and generative AI has more than tripled, up 220%, from 7.0% to 22.4%.

Taking all this into consideration, the same survey notes that "not one marketing-technology activity scored above a 5 on a 7-point performance scale, and those scores haven't moved in two years," he said. "It’s a plateau where adoption surged but results flatlined."

Willis said many companies Domo works with have made "a lot of investments" in AI.

"Companies are spending because they do not want to miss out, despite not knowing exactly how it will all turn out," Willis said. "Two things are critical."

First, the way companies have innovated and solved problems in the past in many ways have been short-circuited by some of the AI tools they use today.

Many of them have created vibe-coded apps to solve problems. It's easy for anyone to be a vibe coder, a software development practice assisted by AI where the software developer describes a project or task in a prompt and the technology writes it. 

"I think for many people, AI has turned them into a server of their ideas, or it has created forgeries of their ideas, because they have not had to do the real work to determine if the idea will even work," Willis said, adding that companies look for that return in investments despite not being part of the original investment

When creating an app in the past, it took an investment in resources and engineering time. They started with the problem, planned, and then prioritized. Then make a prototype and test.

Sometimes with these new tools, people skip the first three steps and go directly to the prototype, he said, adding that he sees it happening with companies all the time. They skip the work that would prove the app could solve their problem.

“In many cases organizations create plausible solutions, and it looks like they should work,” he said. “Some call this a form of AI theater.”

Second, the company’s data foundation creates challenges, which Willis called “the biggie.” When data is scattered people cannot validate it, so the data becomes unusable, despite the investment.

“In many ways it’s like the PC era,” Willis said. “PCs on desktop was supposed to make everyone’s life more efficient. Turned out that didn’t happen.”

The Solo Paradox, named after one of the economists who determined this, emerged. They found that 10 years after the PC revolution, productivity fell by half to two-thirds. Willis believes the same thing is happening here.

“The software didn’t exist to solve the problems, people didn’t know how to use it, and there were different infrastructure and support required — basic experiences that had not happened,” he said. “We are seeing a lot of the same things, but it’s a bit more complicated because the models are remarkable in some ways, but in other they are black boxes, and you’re never quite sure where the failure will happen.”