AI Fever Runs High, But Bad Data Still Costs Big

Tech giants might tout the mind-blowing potential of AI, but in the trenches of the corporate world, there's a nagging problem. Many companies that are bullish on AI are flushing millions down the drain because of shoddy data practices.

It's a paradox that the recent “Fivetran + Vanson Bourne report: AI in 2024” highlighted in numbers. Conducted by independent market research specialist Vanson Bourne, the online survey polled 550 respondents from organizations with 500 or more employees in the U.S., U.K., Ireland, France, and Germany.

The report found that nearly nine in ten organizations are using AI/ML methodologies to build models for autonomous decision-making, and 97% are investing in generative AI in the next 1-2 years.

Yet, trouble is brewing underneath this cheery veneer. A whopping 81% of organizations may trust their AI's results despite "fundamental data inefficiencies." As a result, these companies are bleeding 6% of their annual global revenues (an average of USD406 million annually) due to their wonky AI models.

"The rapid uptake of generative AI reflects widespread optimism and confidence within organizations, but under the surface, basic data issues are still prevalent, which are holding organizations back from realizing their full potential," says Taylor Brown, co-founder and chief operating officer at Fivetran.

A difference of opinion

Curiously, the AI story gets a different spin whether you're down in the weeds or up in the C-suite. Only about a quarter of companies admit they've got full-fledged AI humming along nicely. But, the technical executives who get their hands dirty building and operating AI models daily are way more skeptical of their companies' AI maturity.

Only 22% of technical executives described their AI initiatives as "advanced," compared to 30% of non-technical workers. Regarding GenAI, non-technical workers' high level of confidence is coupled with more trust, with 63% fully trusting it, compared to 42% of technical executives.

There are also gaps when it comes to seniority among technical executives. While those working in more junior positions see outdated IT infrastructures as the top barrier preventing the building of good AI models (49%), their more senior colleagues are pointing the finger back, saying the problem is primarily employees with the right skills focusing on other projects (51%).

What's going on? C-suite types think it's just old IT holding them back. Junior data workers say the problems are legacy infrastructure and technical debt. Their senior colleagues, however, argue they just don't have enough people with the right skills to get the job done. The reality might be a mix of both.

While these high-level strategy debates rage on, a nasty truth remains—data scientists spend most of their time wrestling with messed-up data rather than building those fancy AI models. The survey puts a number on it: 67% of that precious expertise is wasted on cleanup duty, such as cleaning data and fixing broken pipelines.

It always comes down to data

The root of the problems facing underperforming AI programs is the same. The report points to inaccessible, unreliable, and incorrect data as the primary culprit. Most companies struggle to access all the data needed to run AI programs (69%) and cleanse the data into a usable format (68%).

GenAI complicates the situation further. 42% of respondents experience data hallucinations, which can lead to ill-informed decisions, reduce trust in LLMs or the willingness of staff to use the tool, and consume staff time in locating and correcting the data.

This hallucination worry is about to hit the boardroom. 60% of senior management use GenAI. Since they're responsible for strategic decisions, any issues with the quality and trustworthiness of data will be further amplified.

Get ready for the data governance boom

This might be the silver lining: companies are investing in AI's less-sexy parts. They're looking to fix the messy data flow and establish good governance.

In the report, "maintaining data governance" and "financial risk due to the sensitivity of data" are tied to the top spot of concerns among companies (37%). 67% of respondents also plan to deploy new technology to “strengthen basic data movement, governance and security functions.”

Will it be enough to bridge the gap between AI dreams and AI reality? Only time will tell.

Image credit: iStockphoto/photoschmidt