Everyone and their cat is talking about AI. If you haven’t had a “you try out ChatGPT yet?” discussion with the postal carrier, dentist, grocery store cashier, great aunt, plumber, doctor, or anyone with whom you’re likely to engage in some sort of small talk, you probably will soon.
And rightfully so—we know that ChatGPT and the foundational technology it’s built upon, large language models, will have a transformative impact on data and analytics.
So, what are all these people in the biz saying about it exactly? Is it love? Is it hate? Is it just indifference? 🤷♀️
Let’s take a look back at the recent Gartner Data & Analytics Summit—which gathered more than 4,000 data and analytics stakeholders (with more than half at a manager level and above), and included more than 220 sessions and keynotes on the latest data technologies and trends (with more than one quarter of them labeled by Gartner as an “artificial intelligence” topic).
What better place to explore hot takes on AI? From sessions to round table discussions to casual conversations at the booth, lunch pavilion, and dessert tables, here are 10 notable things we heard.
In This Post
- 1) The rise of generative AI = as big as the internet era
- 2) No, AI isn’t going to systematically replace humans.
- 3) But you're not alone if you have concerns.
- 4) AI-driven intelligence is crucial in a post-Covid world.
- 5) AI is taking self-service analytics to a whole new level.
- 6) You might actually be hallucinating.
- 7) Look at potential AI misuse by students in a different light.
- 8) Organizations are harnessing AI to practice sustainability.
- 9) This could be the rise of the prompt engineer.
- 10) Be vigilant.
- Learn more
1) The rise of generative AI = as big as the internet era
Here’s a bold statement to kick things off:
“Generative AI is a business game-changer. It will be as transformative as the internet, and we will experience similar highs and lows. Stay informed. Stay skeptical.”
That’s how Kurt Schlegel, research vice president at Gartner, described generative AI during the first day’s opening keynote, setting a precedent for the rest of the week in terms of what to expect from GPT chatter.
Just for fun, we also noted how long it took for GPT to be mentioned in this 45-minute opener—15 minutes. 🙂
2) No, AI isn’t going to systematically replace humans.
Will we reach a point where AI will be smarter than a person?
To answer that question, during his “The Future of AI” session, Whit Andrews, distinguished VP analyst at Gartner, reminded attendees of the sheer massivity of the world. To figure out if AI is smarter than every human, we’d have to pick from one of the 7 billion people on earth, he said. And we can’t—and don’t really want to—do that. Augmenting—rather than replacing—human capabilities, AI workflows still need some sort of human validation.
“AI is a journey apart from human intelligence,” he explained. “We aren’t waiting for AI to replace humans. We’re investing in applications of technology that allow us to make a more compelling digital parallel to human intelligence.”
3) But you're not alone if you have concerns.
Piggybacking off the previous point, Andrews pointed out a Gartner study conducted to gauge consumer and employee attitudes toward AI.
According to the survey results, approximately one in four people is “AI-friendly”: i.e., eager for AI to be employed at their company, and one in four is “AI-averse,” in that they think it’ll make the world a “less equal place.” The other half is “AI-conditional,” for which Andrews used this analogy: They’re okay with a doctor having access to their data, but not the government. It all depends on who’s using AI and for what purpose.
4) AI-driven intelligence is crucial in a post-Covid world.
As organizations grapple with things like inflation, rising costs, and changing consumer behavior, data is more critical than ever.
In her session with Tellius, Jaime Romano, Haleon’s senior director of category development and capability/NRM, stressed the importance of real-time data and analytics to make decisions in a precarious macroeconomic environment.
“I used to know what size I would need to promote at what point in the season—who’s buying it, and where they’re buying it,” Jaime said. But Covid dramatically changed consumer behavior, forcing companies to lean on data-driven insights rather than instinct or experience.
That’s where AI comes in—Haleon, the world’s biggest consumer healthcare company, is leaning on AI-powered decision intelligence to increase brand growth and drive efficiency and agility.
“Real-time data and analytics is more important than ever,” she said.
5) AI is taking self-service analytics to a whole new level.
As the definition of self-service analytics has changed over the last decade, so too have its capabilities, enabled largely by AI.
This sentiment emerged loud and clear at a lively roundtable discussion we hosted with over 20 D&A leaders on the topic of self-service analytics—specifically, how self-service analytics can be taken to the next level using AI and ML. The first question at the roundtable, “What is the definition of self-service analytics?,” prompted a range of answers. In an ideal end state, one participant mentioned the idea of providing business users with an experience like Siri or Iron Man’s J.A.R.V.I.S.: i.e., a natural language interface for an advanced AI system, specifically for analytics.
Self-service analytics is entering a new era where AI and machine learning are enabling capabilities beyond just on-demand reporting and visualization. With AI, businesses are increasingly gaining a faster and easier path to data-driven insights, and we’ve reached a point where self-service analytics is becoming accessible to everyone in your organization.
6) You might actually be hallucinating.
During “The Enterprise Implications of ChatGPT and Generative AI” session, Arun Chandrasekaran, distinguished VP analyst at Gartner, discussed the propensity LLMs have to hallucinate, or to “make up facts,” he said.
Hallucinations could be present anywhere from 10% to 20% of the time with GPT, depending on the domain, Chandrasekaran noted.
“Arguably, they hallucinate more with logical reasoning or quantitative reasoning tasks,” he said. “And they hallucinate less in domains like, for example, software engineering.”
Noting other risks presented by LLMs—copyright issues and potential for misuse, for example—Chandrasekaran concluded that these challenges are generally “nontrivial” for enterprise deployments, although the onus is on the industry to continue to mitigate these risks.
7) Look at potential AI misuse by students in a different light.
In speaking with an analytics leader at a university, we learned that there are, indeed, legitimate fears about students who are misusing generative AI: e.g., writing entire papers using ChatGPT outputs. One way teachers are thwarting this is by assigning papers related to topics that they know ChatGPT isn’t yet trained on.
Here’s what Daryl Plummer, VP and distinguished analyst at Gartner, had to say about the quandary of AI-written assignments:
“Rather than worry if ChatGPT is doing your students’ homework, worry instead about teaching students to be critical thinkers and evaluators of ChatGPT outputs.”
8) Organizations are harnessing AI to practice sustainability.
Here’s a possibly lesser-thought-about but significant use case for AI: sustainability initiatives.
Lydia Ferguson, senior director analyst at Gartner, laid out several ways in which data & analytics and AI are helping to power a greener future:
- Vegetation and farming: AI is used to produce aerial and satellite images, and this data is used for wildfire prevention and vegetation management.
- Logistics: One of AI’s “original” use cases is for route optimization, transportation, and mobility, increasingly being improved by D&A and AI: e.g., accurately forecasting wind patterns for airlines to optimize routes.
- Waste management: Prescriptive analytics and knowledge graphs can help track movement of waste materials, recommend targeted treatment, reduce unnecessary shipping, and boost the reuse and reprocessing of these materials.
Based on the growing popularity of these types of initiatives, she said, “Roles such as machine learning engineers are really even starting to be more in demand than a data scientist.”
Keep in mind, though, that the usage of AI at scale means more energy usage, thus increasing the overall carbon footprint. To combat this challenge, Ferguson suggested prioritizing cloud data centers powered by renewable energy, tracking energy usage as a key AI metric, disposing of unneeded data, avoiding data duplication, and applying emerging practices for more energy-efficient ML.
9) This could be the rise of the prompt engineer.
During the final day’s keynote, “The New Economics of Technology and Data,” Gartner’s Plummer pointed out that if you don’t know how to ask good questions, you can’t learn anything, essentially. With that, he predicted that prompt engineering will be the “coolest skill” in D&A within the next few years.
By this, he meant that business and data analysts will learn to maximize the outputs of their firm’s investments in generative AI capabilities, while in the meantime, a class of ML engineers and data scientists will emerge who actually tinker with the LLMs and underlying algorithms. So, hey, your ChatGPT tinkerings of today could actually be the coolest skill in D&A tomorrow. 🧚🏻
On the same note, don’t forget about the rise of the analytics engineer over the past couple of years—i.e., an engineer typically between a data engineer and data analyst who is upskilled to do transformations with tools like dbt, for example.
10) Be vigilant.
Every analytics leader should explore the vast possibilities of pairing generative AI with analytics. But we’ll leave this list with another remark from Andrews’ “The Future of AI” session:
“Powering AI with less data and fewer experts will make it cheaper, easier, and riskier.”
He recommended, among other things, leveraging commoditized AI (vendor-supplied and logic-based) to reduce time to delivery; providing the right training for your teams; and, importantly, governing AI with a central team so you can properly set priorities, fund initiatives, and measure success.
On implementing AI at your organization, Andrews added, “The most important investment you can make is on governance.”
When it comes to generating precise analytical answers to your business queries, it pays to work with an established AI-powered analytics vendor who has put the proper guardrails and governance in place to join the power of generative AI with augmented analytics. Learn more about our new Tellius Copilot feature, which takes our platform’s augmented analytics capabilities to the next level by leveraging the power of GPT.