Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Nearly all data and IT pros use AI, but few are heavy users.
- Many would give AI agents unrestricted data access.
- AI data prep and validation take about 10 hours a week.
If you’re curious about what’s happening in the eye of the artificial intelligence storm, look no further than what the data analysts of the world are up to. They’re bullish on AI, of course, but they’re still using spreadsheets, and barely a handful are working with real-time data.
That’s the word from a new global survey of 700 data analysts and 700 IT leaders from Alteryx. While 96% report using AI for their work, only half can be considered frequent users of AI tools — 49% report they use AI always or most of the time.
Also: 51% of professionals say AI workslop lowers their productivity – stop it in 2 steps
Agentic AI is high on the agenda, with close to six in 10 respondents, or 59%, predicting they will be actively employing AI agents within the next 12 months. In addition, at least half say they are willing to grant AI agents “unrestricted access” to their data.
The security implications of such access were not discussed in the survey report, but 44% did specify that it was critical to include human oversight as part of such access.
The most common agentic AI applications
The most common agentic AI applications now in production are drafting communications and scheduling workflows.
Where AI agents are being put to work:
- Drafting standardized communications or summaries for stakeholders: 59%
- Scheduling or routing workflow tasks, such as alert triage and process automation: 54%
- Generating standard reports or dashboards without manual intervention: 48%
- Monitoring key performance indicators and triggering alerts or actions: 45%
- Cleaning, preprocessing, or validating routine data sets: 45%
- Running routine statistical analyses or basic predictive models: 34%
- Automatically generating insights or recommendations from data: 23%
“Foundational data work” — cleaning and prepping data for ingestion by AI models or associated retrieval-augmented generation platforms — still takes up a chunk of data analysts’ time. Respondents report spending close to six hours per week on such tasks, with 48% spending six to 10 hours weekly. The tools they use to handle such work are spreadsheets, cited by 61%, followed by business intelligence tools, cited by 56%, and dedicated data preparation platforms, as indicated by 51%.
Also: Building an agentic AI strategy that pays off – without risking business failure
“The continued dominance of spreadsheets reflects a broader reality,” the survey report’s authors suggest. “AI is layering on top of existing workflows rather than replacing them.”
Another surprising finding is that despite all the attention to real-time responsiveness, few organizations truly have real-time capabilities. Only 20% report that moving from data analysis to a business decision can be done within a few hours, and a mere 5% say they support real-time decision-making.
The biggest barrier to AI?
Explaining AI outputs to business decision-makers, the respondents say. There is also a notable lack of analytical skills across businesses.
Barriers to AI in business decisions:
- Difficulty interpreting or explaining AI outputs to decision-makers: 55%
- Limited analytical skills among business users: 54%
- Data is not sufficiently clean, integrated, or governed: 50%
- Lack of clarity on ownership or accountability for decisions: 49%
- Technical limitations of AI tools or infrastructure: 45%
Generating insights from AI is not a once-and-done exercise by any means, and it also gobbles up more of data analysts’ time. The analysts in the survey spend almost four hours per week validating or correcting AI-generated outputs. One in six say they spend almost an entire workday, six hours or more, fiddling with AI results. Add the six hours spent on foundational data work, cited above, and this adds an AI “tax” of almost two days per week to professionals’ time.
Also: Over 80% of US government agencies already use AI agents – and it’s only the beginning
This points to an emerging skill set that is becoming more valuable in the AI age: validating AI outputs. This is “a signal that while AI can accelerate work, organizations still need human oversight to ensure outcomes are consistent, explainable, and trusted,” according to the survey’s authors.