Advanced artificial intelligence, such as ChatGPT, is said to be the most disruptive technology since the advent of electricity. The technology, known as generative AI, can produce content, including text, images, and audio.
“We’re seeing an impact that’s largely unprecedented,” said Greg Coulombe, Director, Emerging Technology for Intuit Canada. “AI and generative AI, specifically, is reaching every part of our lives.”
The new tools are so revolutionary because they open the door for anyone to use artificial intelligence. “It used to be that people literally had to have a PhD in math to use AI,” said Coulombe. “Now we have democratized the ability to do very powerful things.” He predicts this will lead to an “explosion of ideas and innovation” in the next few years.
With all the hype, how can businesses separate fact from fiction? Coulombe shared five tips to help organizations keep pace with the evolving world of AI.
Start experimenting now
Coulombe expects that generative AI is here to stay. “It’s no ‘flash in the pan,’“ he said. “The best thing you can do is start using it. You can’t learn how to ride a bike by watching someone ride a bike. You have to get on the bike and pedal.”
Generative AI is moving so fast that companies can’t afford to sit on the sidelines. New applications that generate business value are being introduced every day.
Every technical or business person should be learning about generative AI technologies, such as Chat GPT, said Coulombe. “It’s cheap and easy to learn. What’s more, “you may get inspired and help your company do amazing things that no one thought was possible,” he said.
It’s not ready for critical workloads
While Coulombe thinks people should learn the new tools quickly, he does not advise them to start deploying highly important production workloads on AI yet.
“The tools can do some shockingly amazing things, but they are not intelligent or sentient,” he said. “Quite simply, they are a statistical model that, when given a piece of text, knows how to predict what text comes next.”
There have been challenges with the tools where they “make up things in a very convincing way,” Coulombe explained. This is called “hallucination.” As well, they can spout text that could be objectionable or hateful.
Coulombe notes that generative AI isn’t quite ready for mission-critical industries such as healthcare, and other sectors where accuracy is paramount.
“At this point, this technology would not be something you’d want to use and expect it to magically do your taxes,” he said. “There’s just too much risk of something being incorrect or hallucinatory.”
Where the business value is now
Generative AI is not new, notes Coulombe. Intuit has been using it for many years to summarize conversations between a user and a tax expert, resulting in the elimination of more than one million hours per year from customer call times. As well, the company is using Chat GPT to help small businesses reduce the time it takes to write advertising or email copy. However, Coulombe notes that the generated copy must always be reviewed by the business owner.
“These are great examples of how we’re using AI to function as a decision support system, helping both customers and our experts make insightful financial decisions,” noted Coulombe.
Small businesses new to AI may get the best bang for the buck by investing in more “old school” fundamentals, suggested Coulombe. This includes tools to predict cash flow or to classify data, such as bookkeeping line items. “This is where you’re going to most directly impact your business operation,” he said.
Knowledge engineering is another key use case to consider. This is a branch of “classical” AI that establishes the core business logic and rules for services. “It’s unlocked one of our most game-changing innovations,” said Coulombe. Intuit’s “Explain Why” feature uses the knowledge to calculate taxes, and explains why there’s a balance owing, for example. “In the future, it could be paired up with Chat GPT to provide advice on how you could reduce your taxes next year. We’ve barely even scratched the surface of what AI can do,” said Coulombe.
Prepare and protect your data
Most organizations have messy data, said Coulombe. “Pulling your data together and organizing it is an amazing first step, and a prerequisite to using AI powerfully,” he said.
At the same time, organizations should establish principles to use AI responsibly. “Be deliberate about how you intend to use AI,” Coulombe said. “You want to think about how it can benefit your business while keeping your customers’ best interests at the center.”
Policies on ethics and privacy are critical. To use generative models, organizations must send their data to third parties for processing. “That’s a massive gotcha,” said Coulombe. “At Intuit, customers’ data and privacy is absolutely paramount.” All organizations need strong rules on how to use these services while respecting the privacy and security of customer data, he stressed.
Playing with the technology is the best way to learn about it, said Coulombe.
“Ultimately, you’re still in control,” he said. “The goal here is never to replace humans. It’s to augment them. AI and humans together are more powerful than either of them separately. That’s where we see the biggest business value.”