Three hot AI debates that you need to have on your radar

When a new technology first arrives there are always unknowns, but typically the fundamentals aren’t in question.

Take cloud computing, for example. There were discussions around security and cost-effectiveness but the basic proposition—what it was, how it worked, what it could do—was clear.

With generative AI (gen-AI) it’s different. The science is not settled, the best practice evolving, and the potential impact either world-changingly transformative, or strangely limited, depending on who you speak to.

It is, in fact, a field awash with debates, which is why IBM and WIRED Consulting are partnering on a debate series that will explore the future of the technology that continues to dominate the headlines.

IBM’s CEO in the UK, Leon Butler, explains that hosting frank conversations about AI will help business leaders to cut through the clutter. “It’s hard for leaders to make business investment decisions with confidence when dealing with complexity and change,” he says. “We want to help leaders to step back and learn from experts who can guide the decisions they face.”

Here are three conversations that business leaders should have on their radars…

Is generative AI overhyped?

Just how significant will the impact of gen-AI prove to be? The widely articulated base case is that it will reorient whole industries. Some commentators go further, suggesting it will be more transformative to human existence than electricity.

In light of such beliefs, companies are pouring money into gen-AI. This year alone, Meta, Amazon, Alphabet, and Microsoft are set to to spend over $300 billion between them, such are the costs involved in training and running frontier models. Investors seem to agree that the juice is worth the squeeze. OpenAI is currently valued at $260 billion—around ten times what it was worth in 2023.

And companies are planning to re-train staff at scale. In a recent global study by IBM’s Institute for Business Value, surveyed executives estimated that implementing AI and automation will require 40 percent of their workforce to re-skill over the next three years. IBM is lining up resources to support this with a commitment to skill 30 million people globally by 2030.

Those looking for concrete results of all this investment can point to the continuous breakthroughs in multimodal AI, the vast volumes of traffic flowing to gen-AI chatbots, and the increasing level of enterprise adoption, particularly of tools built on top of foundation models.

Proponents highlight the way it is transforming workflows for software developers, marketing teams, creatives, customer service operatives and others. IBM, for example, says that AI and automation is on track to deliver $4.5bn of productivity gains across the company by the end of 2025. This is a result of tools such as AskHR—an HR assistant underpinned by gen-AI—which lets managers complete HR transactions 75 percent faster than before.

But although companies are driving gains with gen-AI, sceptics question whether these are at a level that justifies the towering hype. They point out that alongside the success stories, many companies are struggling to realise ROI from their deployments. A recent McKinsey survey, for instance, showed that 78 percent of organisations use AI in at least one business function—yet more than 80 percent of those companies aren’t seeing a tangible business impact.

Moreover, they say, gen-AI’s use cases and capabilities are heavily limited by its error rate. There are plenty of tasks where, if the output is even 1 percent inaccurate, it’s 100 percent useless. This makes the notion that it’s the new electricity seem like a stretch.

So where does this leave business leaders?

A take that resonates with us is that even if generative AI doesn’t radically reconfigure the economy, there are still plenty of companies using it right now to improve the bottom line. An analogy might be drawn with the dotcom bubble: when it popped it was clear that a lot of the frenzy was hot air. But it didn’t mean the internet wasn’t real—and many of the companies that survived went on to transform our world.

Understanding where gen-AI can deliver proven value, and tracking how that envelope shifts as the technology matures, will be essential.

Will gen-AI progress hit a ceiling?

Developments in gen-AI have progressed speedily over a short period of time. It was only November 2022 when OpenAI unveiled its trial version of ChatGPT, saying it was “excited to get users’ feedback and learn about its strengths and weaknesses”. Since then, frontier models have proliferated, and they have continually leapfrogged each others’ benchmark scores.

As they have improved, they have put new capabilities in the hands of enterprise. Recently, for example, radical improvements in video models have changed what it means to produce video content, and state-of-the-art coding models have done the same for software engineering.

But can this rate of progress continue? That’s an important question for business leaders, and their take on that will inform everything from product and innovation roadmaps to M&A decisions and workforce strategy.

Some believe the trend can’t hold. They point to indications from frontier labs that the original scaling law of ‘more data and more compute equals better performance’ is producing diminishing returns. And they question whether the recent approach of putting more compute into the ‘inference stage’—the phase where the model produces its output—can continue to drive progress. A recent study shows that the benefits of this approach tapers off as tasks become more complex. Moreover, does common sense not dictate that there are, ultimately, limits to data availability, computing resources, and model architectures?

There is an alternative framing that, in a sense, accommodates this range of possibilities. Even if the current generation of models are as good as it gets, the argument runs, their potential has barely been tapped.

Not only have many businesses not yet taken advantage of the more obvious tools and applications, but, as many point out, we still haven’t figured out how to make best use of what the models themselves have to offer. There remain many years of development work and creativity ahead as we figure out how to take full advantage of their capabilities.

“Barely one percent of the world’s enterprise data has been consumed into LLMs,” says IBM’s Leon Butler. “And, with open-sourced smaller models championed by IBM, the cost of deployment and scaling will continue to drop, and so give birth to more companies like DeepSeek. There is still a lot of upside to extract.”

Will AGI happen sooner than we expect?

Most experts agree that it is a question of when, rather than if, we achieve artificial general intelligence (AGI). It is famously a term whose definition keeps shifting, but in common parlance it tends to refer to a form of AI that is able to match human capabilities in every task.

This is a separate conversation to whether gen-AI can keep improving. Gen-AI could theoretically make dramatic leaps while still remaining ‘narrow AI’, for example, highly capable but only within a defined scope. Equally, gen-AI could plateau without negating the possibility of AGI, since progress could come from other regions of the AI landscape.

So, will AGI emerge and, if so, when?

For business leaders this isn’t a purely philosophical question, there’s also a practical implication for workforce strategy. Right now, AI’s limitations mean it is generally used as a tool to augment rather than replace human labour. But if AGI is as capable as a human, that may no longer hold true.

Perhaps AGI will give us firms where much of the workforce is AI. Or perhaps it will lead to a future where humans are still widely employed, but just in specific capacities: defining the problems AI works on, ensuring it adheres to defined values, and bringing a strain of higher-order creativity to the table that may continue to elude machine learning. Human employment may also shift to roles that require embodiment.

Until recently, surveys of researchers tended to predict that AGI would arrive around 2060. However, with advancements in large language models (LLMs), and a belief that the transformer model architecture that underlies them could provide a pathway to AGI, that date has pushed forward. In 2023, a survey of experts predicted AGI by 2040. More recently, leading players in AI have suggested we could see AGI before the end of this decade.

The recently published AI 2027 scenario, written by a group of expert authors, vividly describes how current trends in AI development mean AGI could be with us far sooner than we ever imagined. In this version of the future, AI models train ever more powerful AI models, and within just a few years, superintelligent AI far outstrips human geniuses.

However, versions of the future also exist in which progress is far slower, and perhaps never leads to AGI. A recent research paper suggested, at the very least, a less accelerated progression. It looked at reasoning models, an enhanced version of large language models that are trained to solve multi-step reasoning tasks, and found that they had “fundamental limitations”.

When presented with highly complex problems, the models faced what the researchers termed a “complete accuracy collapse”. Not only this, but as the models neared collapse, they began “reducing their reasoning effort”. With the air of disappointed school teachers, the researchers said they found this “particularly concerning”. And that’s putting aside the fact that frontier models still struggle with basic maths among other seemingly elementary tasks.

The typical response when an organisation is faced with an uncertain future is scenario planning. That might seem irrelevant in the face of such a paradigm shift as AGI, but there’s an argument that this could still be worthwhile.

It’s arguably unlikely that AGI—if it were ever achieved—would suddenly be everywhere all at once. Is it not more realistic to think it will be implemented in different ways in different places at different speeds? That would make it possible to position a business strategically to take advantage.

To continue that thought experiment, you might ask yourself: where does your organisation rely on having the smartest people? If intelligence becomes a commodity, that may no longer be a competitive differentiator. If you can shore up your moat in other ways, that could be prudent and could strengthen your position. With or without the advent of AGI.