Three questions business leaders need to ask to get the most out of generative AI

Pedro Arellano Profile picture for user Pedro Arellano Salesforce May 9, 2023
How to bridge the gap between admiration and implementation of AI? Pedro Arellano has three priorities to guide you.

Bot with speech bubble, concept of chatting with bot. Asking question to AI. Question mark and bot representation ready to work © SaskiaAcht - Shutterstock
(© SaskiaAcht - Shutterstock)

Have you heard? We’re living in the era of AI.

As I write this, people are using generative AI and large language models (LLM) to do everything from build a website to write a customized resume. This explosion of use means we’re living through two incredible technical achievements: first, people with non-technical skills are becoming more comfortable completing tasks that once required unique knowledge and training. And second, a revolutionary shift in roles is on the horizon, especially in the world of analytics. We know that AI can augment human tasks, but humans will also augment AI by becoming more strategic and leaving the raw data analysis to the machines. And while today's question is what are these tools capable of, tomorrow's question is how can they transform everything we do.

To get to that future state, however, businesses need to implement AI into their processes - and they aren’t there yet. In this March 2023 Salesforce survey, senior IT leaders reported seeing the value in using generative AI in their business, and even the skeptics who say it’s ‘over-hyped’ believe the technology has the potential to help them better serve their customers, take advantage of data, and operate more efficiently. But there are conflicting thoughts on if – and how – to implement AI within their business. In that same survey, 65% of senior IT leaders said they can’t justify the implementation in their business right now, while a recent Gartner poll shows a desire amongst executive leaders to increase AI investments.

Why the confusion on if, when, and how to bring in AI? Implementing generative AI requires a solid understanding of a company’s data and a solid enterprise data strategy. Business leaders need to know where their data lives, how it’s collected and cleaned, who has access to it, and how it’s analyzed. That may sound easy, but recent research shows 41% of business leaders lack an understanding of their data because it is too complex or not accessible, and 33% lack the ability to generate insights from it.

To bridge the gap between admiration and implementation, companies should be considering three data and AI priorities:

Priority #1 – make sure you understand what generative AI is good at… and what it’s not

LLMs are amazing at generating, summarizing, and predicting language, but they're not so good at math. This means they can’t replace analytics engines, but they can work together WITH an analytics engine (like Tableau) to deliver experiences that help businesses – and workers – make more impactful decisions faster. These experiences include more conversational interfaces where analysis feels more like question and answer, anticipating questions people might ask, suggesting questions they might not have thought of otherwise, or communicating large collections of insights in easy-to-understand summaries.

Priority #2 – lead with governance and responsible applications of AI

Delivering trusted and ethical experiences will be key to success with generative AI investments. Developing a curated semantic foundation will give LLMs the business context they need to deliver trusted and meaningful insights while eliminating the risk of hallucinations. This is especially true when it comes to analytical roles; data analysts who can blend raw data with the right business context will become even more valuable because they’ll make AI-generated insights significantly more trusted for the people that consume them.

Priority #3 – make sure humans will augment your AI

People have to remain involved in key elements of the analytics process including curation, governance, modeling, validation, and objective-setting. However, the benefit of AI is it will allow many to become increasingly passive in their day-to-day engagement with the analytics system and leave that work to the machines. Analysts with strong data skills will still be needed (in fact, they become essential to the success of generative AI for analytics), but the population at large will be able to interact with the AI very naturally and without a data skill requirement.

AI is here to stay – but it's only as good as the data it's given, and that data is only as good as the skills possessed by the person using it. Companies that hire or train employees with the right data literacy skills will find it easier to fuse AI into their daily operations. And a successful data strategy in the age of AI will recognize the changing roles of people to maximize the potential of this revolutionary technology.

A grey colored placeholder image