Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Another AI winter could be coming, but that’s when the real work heats up

While many of the emerging AI iterations can rightfully be seen as “glimpses into our future,” they are not necessarily business ideas. And therein lies the rub. 

Employees hard at work

[Images: Adobe Stock/Svitlana]

Profile photo of Ryan McDonald

BY Ryan McDonald

Artificial intelligence (AI) is writing poetry, drawing pictures, and taking over search queries. It seems every day, I read about another ChatGPT or DALL-E iteration being rolled out from open-source platforms. And Google and Adobe just teamed up to bolster Google’s Bard chatbot. The AI excitement is palpable, and so are the investments. According to a forecast from the International Data Corporation, “global spending on artificial intelligence, including software, hardware, and services for AI-centric systems, will reach $154 billion in 2023.”

While many of the emerging AI iterations can rightfully be seen as “glimpses into our future,” they are not necessarily business ideas. And therein lies the rub for the current AI enthusiasm. 

I am not an AI naysayer; I believe it will dramatically reshape our lives (and already has to an extent). But I have a practical view on the subject due to two decades of being around this technology, seeing AI models go from academic concepts to useable applications and potentially transformational ones.

During my undergraduate years at the University of Toronto, I assisted Professor of Computer Science Gerald Penn, who was working on a project to summarize news articles for mobile devices—it was in the year 2000! I loved it. It was clear to me back then how hard it was for computers to understand even simple things about language, but it’s long been clear to me that AI has profound possibilities.

Now, if you had told me 10 years ago that AI or machine learning could write a poem in 2023, I wouldn’t have believed you. And I work at a company that offers AI-driven customer service chatbots and digital assistants to help humans do more work more easily. 

So, I am pro-AI but also pragmatic about entrepreneurs’ prospects in turning machine learning and data into a goldmine. To keep the investments and buzz alive, the tech community should produce use cases that make work and living a better experience. It will take a ton of work to make that leap from simple use cases to viable startup businesses that handle complex tasks. 

Before we get there, I believe there is a serious risk that tech will see an AI winter. Funding for AI startups—which was slowing down last summer before seeing a post-ChatGPT rush in recent months—will again flow less abundantly if not freeze altogether. It’s inevitable for a pair of key reasons.

HISTORY TELLS US SO

First, let’s look at the last half-century. There was an AI winter as far back as the 1970s and early 1980s when early optimism about AI research was not matched by results, leading to a decline in funding and interest. 

It happened again in the late 1980s and 1990s due to the over-hyped expert systems and other AI technologies, which ultimately failed to live up to lofty expectations. Another AI winter occurred in the late 1990s and early 2000s when the dot-com bubble burst, causing a larger decline in tech funding, including for AI research. So, we’ve seen this cycle before with AI and other emerging tech, such as Internet of Things, which was a digital transformation topic du jour in 2015 before settling into the background. IoT’s up-and-down trajectory has happened even as developers continue to faithfully code and experiment with smart products—some that millions of us will use in the future if we aren’t already. AI will likely have similar ups and downs ahead.

INVESTORS DEMAND MORE THAN EXPERIMENTATION

In 2023, the viability of use cases is still uncertain, and AI startups will have to show ROI to get more funding in an economy with rising interest rates and persistent headwinds. They have to become businesses that make money. 

To predict the use cases that will matter, let’s look at the last decade of tech. We’ve figured out how to categorize email spam with machine learning. Then we progressed to digital assistants. Both are narrow use cases with narrow complexities—that’s why they’ve already been accomplished. Today, we are seeing AI speeding up workflows with ChatGPT, and in the next few years, emails won’t just be augmented by recommended phrases and short sentences—they will, at times, be wholly written by AI. 

Other daily activities will become automated, too. With a lot of work, these applications can build strong and promising companies like the one I currently work for.

But the current experimental activity in the space means droves of startups eventually fall by the wayside. Every day, there’s a new startup that’s going to do something with ChatGPT, but many of their founders will learn they underestimated all the work needed to build a real company around their idea. This last mile of product development is often harder than building AI models. The user experience, latency, cost, etc., are all critical to elevating an AI model to a usable product that adds value and ROI. 

What use cases can become viable businesses? I predict they’ll come from industries such as manufacturing (autonomous cars and IoT devices that work as advertised), climate (smarter energy consumption), medicine (better health treatment and wellness practices), science (deeper understanding of Earth and the cosmos), and customer experience (personalization). AI-powered chatbots already impact areas such as customer service, and it’s just the beginning. 

A REALITY CHECK IS COMING

A new AI winter seems necessary—a reality check—in a Darwinian sense. A thousand flowers are blooming, but only a few will be picked. I believe those that get picked will not necessarily be doing the coolest AI, but they will be focused on the last mile of connecting AI to a business outcome. 

The work that goes into winning AI tech will be immense and impressive. Further, the apps that survive the next AI winter will be a hot investment because they will be used broadly by B2B customers, consumers, or both. I predict their use cases will improve the lives of humans, at first in fairly narrow ways but eventually in complex ones. And for them, the hype will be real and worth the wait.


Ryan McDonald is Chief Scientist at ASAPP.