Artificial intelligence: boom or bust?

When German economist Karl Marx used the term “boom or bust” in the 19th century, it was to explain economic expansions and contractions. Today, the term neatly explains the dilemma faced by emerging technologies such as artificial intelligence (AI).

You don’t have to look further than the recent past for some reminders of the boom-or-bust phenomenon. Remember the dot-com bubble of the late 1990s, when speculation in internet-based businesses saw US equities spike rapidly before the overvalued stock came crashing down? Or the origins of the 2008 global financial crisis, when the failing US housing market ultimately dragged down the global economy?

Right now, the future of AI is poised on a boom-or-bust knife edge. The result of a lineage of technological innovations and machine-learning advances, AI’s credentials should point squarely towards a boom. Yet, for all the hype, tangible applications are still in short supply.

Despite the money being pumped into the development of tools, the big generative AI tools like ChatGPT, Bard and Bing AI are largely being kept busy writing reports, distilling technical information and fielding questions about religion, retail choices, how to stop (or start) a war and marital concerns. This highlights that while people, businesses and society have embraced the promise of AI, we still aren’t quite sure how to leverage it.

This might have something to do with the origins of AI as part of a theoretical, mathematical and computer science universe. People glaze over when you start talking about data in the same way as atoms and amoebas, but data is the basic building block in the AI family tree. While humans have been generating AI-generated data since the 1950s, we’ve only fairly recently developed the computational power to analyse this information at scale, sparking the rise of data analytics and more machine learning advancements.

Therefore, even though it stands on the shoulders of impressive technological advances, AI is still in its infancy. It will, of course, continue to develop as access to data grows and as machines keep learning. This process should not be rushed — particularly as the process of teaching machines moves towards what we call deep learning, or teaching machines to complete human tasks such as driving a car or recognising objects.

Deep learning can be thought of as the postgraduate-type thinking expected from advanced learners who can draw on experiences from parallel processes to arrive at meaningful conclusions. Until OpenAI was formed by Sam Altman, Elon Musk and Peter Thiel in 2015, this deep-learning leap was largely the domain of big tech companies with deep pockets, such as Amazon, Alibaba and Meta.

When OpenAI’s customer-centric ChatGPT launched on November 30 2022, with its easy-to-use interface, it proved a game changer. No longer were humans instructing technology; they were conversing with machines and finding new and innovative ways to use the emerging AI tool.

Such has the level of uptake been over the year since ChatGPT burst onto the scene that Gartner already believes more than 70% of businesses are now exploring the use of generative AI. ChatGPT alone had nearly 39-million monthly active users and 23-million downloads in September 2023.

To further embed the business case for using these new tools, MIT Sloan researchers recently noted that generative AI could “improve a highly skilled worker’s performance by as much as 40% compared with workers who don’t use it”. According to McKinsey & Co, the current crop of AI tools alone could potentially automate jobs to such an extent that they free up 60%-70% of an employee’s time. This is where things get interesting. Not only do generative AI tools have the potential to improve business productivity and efficiency, but they can also give back the most precious of commodities: time.

From a utility value perspective, OpenAI’s ChatGPT has certainly achieved the “stickiness” needed to continue winning over new customers and retaining existing users. This puts AI squarely in boom territory, but front-runners in the industry are already scrutinising the business models of the world’s top companies by market capitalisation and realising that the big money doesn’t come from selling services or time. It comes from selling products. Herein lies the first potential bust.

The “businessification” of AI presents a pivotal choice for developers: pursue AI hardware or adopt a user fee-based model, both risking customer loss. As the consumer base narrows, tech firms must ensure AI delivers human value. Case studies show AI integration in business systems, like Salesforce's AI-driven customer relationship management platform, promising growth and efficiency, yet awaiting substantial data. Giants like IBM and Google dominate, challenging start-ups. This monopolistic trend raises concerns about control and impact on human life. 

As much as the likes of ChatGPT have won hearts and minds, the plethora of AI start-ups needs some big, visible wins that move beyond basic predictive analytics and machine learning and start to tap into deep learning. When we eventually see an AI algorithm capable of spanning industries and applications, being able to fly an aeroplane and then use the same intelligence to drive a car, that’s when we know we’ve hit the boom time of AI superintelligence. Until then, watch out for bubbles.

• About the author: Professor Manoj Chiba presents the online Predictive Analytics Masterclass at the Gordon Institute of Business Science (GIBS). Attend this Masterclass to discover how to leverage analytics for better decision-making and innovation to affect your business's bottom line.

This article was sponsored by GIBS.