The generative AI hype is almost over. What's next?

A recent RAND Corporation report showed that 80 per cent of AI projects fail. That’s twice the failure rate of other information technology projects.

Nonetheless, ChatGPT - the company that kicked off the Generative AI (GenAI) frenzy two years ago - is expected to get a $100bn valuation once it closes its next funding round.

The hype around new technologies usually continues even if they do not deliver on their initial promise, at least up to a point. According to the Gartner hype cycle, inflated expectations will be followed by a trough of disillusionment.

GenAI is probably at this turning point right now, as a Gartner report suggested in June. This does not mean that the advances in Large Language Models (LLMs) have not been real, but it alerts us to the difficulty of translating technology into economic growth engines. We simply expect too much in too soon.

Technology historian Carlota Perez explains that primary technologies - like LLMs - always require a second wave of technology innovation that involves the development of applications and adjustments of organisational structures.

Electricity for example only became impactful, once electric motors were developed and production lines in factories were reorganised to leverage these inventions. Keeping this in mind, companies can adjust their AI adoption strategy.

Using GenAI like Google

What do you do when you are trying to find out the difference between Machine Learning and GenAI? You google it. Google then provides you with a list of links where you can dive into the specifics.

More recently you also get a brief AI generated answer. In most cases this will suffice.

You can also pose the question in a GenAI application to start with. This has the advantage of starting a conversation where you can ask further questions.

Hallucination can be an issue, but for many questions that’s not your primary concern. If it is, you can always dive into the specifics afterwards. Learning is not a linear process anyway.

While using GenAI this way is efficient, the less obvious yet more important benefit is the gradual familiarisation with AI tools. With time you figure out which prompts are more effective and how you can separate fact from fiction with higher accuracy.

You will also learn which tasks these tools are most suitable for. When I asked executives in my MBA class, they name two different types of tasks.

Some use it to replace relatively simple jobs, which previously they outsourced, e.g. helping them draft a press release or a very straight forward legal document (one that is not high-stakes).

Others use it to come up with new ideas, e.g. looking for examples from other industries which faced similar issues.

From an organisation perspective, the wide-spread use of GenAI is a necessary precursor to more ambitious integration of AI into its operations. If people are not comfortable with the technology they will resist.

View AI as a change project

It’s easy to see AI primarily from the technical angle. That is a big mistake.

Adopting AI requires new business processes and new behaviours. Inertia is a strong force which is hard to overcome.

Making people comfortable with a new technology in principle is only the first step. You need a smart transformation plan.

Eric Siegel provides one in The AI Playbook: Mastering the Rare Art of Machine Learning Deployment. Using UPS as an example, his first insight is that big promises usually scare people more than they inspire them.

When Jack Lewis presented a prototype of a system that predicted tomorrow’s deliveries and prescribed more efficient delivery routes for drivers, the executive’s response was “So, are you working on anything important?”

As a result, he scaled down the program to simply distribute packages better between trucks. It may not have been as grand, but it also required less change, making it more attractive to senior management and easier to implement.

Siegel’s second insight is that labelling matters. Your AI initiative will appear less threating when it sounds mundane. Considering that change is scary, that’s a good thing. Save the big words for the presentations once you have won. Initially go with something like “operational improvements”.

The AI hype is ending. AI gains are not

Finally, the benefits of change are best appreciated when leaders experience them. Jack took a sceptical UBS executive on a ride where the route was prescribed by the algorithm.

At one point a counter-intuitive turn was taken, leaving some packages that were close by for later to create a more efficient route. This is when it dawned on the executive how the new system could create big efficiencies.

Even with this smart transformation approach, it took years to fully integrate AI into UBS’s delivery system. Today it is a central component of the company’s overall optimisation system. Each year it saves 185 million miles of driving and $350m in costs.

Companies are already benefitting from from AI in general and GenAI in particular, and will continue to do so.

But like all new technology, this is not a magic formula and the trickiest bit is adjusting operations. The hype may have peaked but if firms adopt the right strategy to incorporating AI into their operations, the gains are only just beginning.

This article was originally published by Forbes.

Further reading:

Working on the jagged frontier: How companies should use generative AI

Pass the IP: How generative AI will reshape intellectual property

Beyond the hype: What managers need to ask before adopting AI tools

Increase the odds of success in digital transformation

Christian Stadler is Professor of Strategic Management and teaches Strategic Advantage and Strategy and Practice for the Executive MBA and Global Online MBA.

Learn more about strategy on the four-day Executive Education course The Strategic Mindset of Leadership at WBS London at The Shard.

Discover more about Strategy and Organisational Change by subscribing to our free Core Insights newsletter.