To Frankfurt for the second in a series of three European roundtables exploring the power, the promise and the potential of artificial intelligence. The Tech Monitor trifecta – put together in association with AMD – is designed to understand the challenges senior technology leaders face and the paths they are charting as they look to adopt perhaps the most transformative technology of the last fifty years.
The series started in London and moves on to Copenhagen after this stop in Germany. And as before, we started the evening by understanding what AI-related issues were front of mind. These included the challenge of data extraction and preparation; the uncertainty caused by embryonic, ill-formed AI regulations; and the difficulty of educating the C-suite on the practical applications of AI while simultaneously playing down the hype that surrounds it.
Here are five takeaways from a night discussing the potential of AI:
1. AI does not lack for use cases
Asked how they were using artificial intelligence in all its forms within their organisations, attendees offered a wide range of applications. These use cases are either at an experimental stage or in full production and include data-driven solutions to prevent customer churn, detect fraud, and identify peaks and troughs in transactions. Elsewhere, a proof of concept is underway designed to support a help desk address the 80% most common customer queries. Others are applying it in an attempt to attract younger customers. Whereas their London counterparts are mainly experimenting with productivity tools, here in Frankfurt there appears to be no lack of AI use cases.
2. Don’t let the AI hype distract you
Notwithstanding the enthusiasm for their own experimentation, delegates warned against getting distracted by the fanfare currently surrounding artificial intelligence, and generative AI (Gen AI) in particular. One noted how every security tool is today “powered by AI”. The use of literal air quotes around the phrase made it clear that this delegate remains sceptical about the claim. Another attendee said, in direct terms, that his task is to work out “how to protect the topic of AI from all the BS out there. I’ve seen it happen with cloud and mobile, and now with AI.” He said his job, as an AI practitioner, was to “under promise and over deliver”, not to get carried away by the hype.
3. Cloud for now. On premise in the future. Possibly.
Despite reservations about the increasing power of the big tech providers, most attendees are willing to put their early experiments in AI – especially Gen AI – in to the cloud. Why? Because it provides the necessary scalability and compute power. For now there appears an unwillingness to commit significant capital to in-house AI hardware. That position may change as organisations weigh up the ongoing cost of hosting, processing and transferring their data in the cloud, a typically expensive activity. Attendees were urged to identify unused capacity in their own data centres. “Review it and optimise it.”
4. Collaboration unlikely to solve the compliance conundrum. Unfortunately.
More than one attendee identified regulation – both existing (think GDPR) and forthcoming (think the AI Act) – as a potential barrier to AI progress. Some feared that projects embarked on today might be stymied by the regulator tomorrow. As a solution to this problem, a “philosophical idea” was presented to all: what if multiple organisations from a variety of vertical markets shared their data and strategic approach with those putting together new directives? This way the regulator would better appreciate the business models of those they seek to guide. Greater understanding and empathy would, in turn, lead to more pragmatic lawmaking. While a worthy idea in practice, attendees offered a handful of practical limitations. These included the “bureaucratic” challenge. In other words, such an exercise would become cumbersome, shapeless and leaderless. One attendee was more sanguine about compliance. “It is the cost of doing business,” he said.
5. Remember, not every problem is a nail
Asked to share advice on embarking on an AI project, attendees offer a range of best practice tips. These included securing your projects from the outset and ensuring that the data that forms the basis of any large language model (LLM) is as “clean” as it can be. The old IT maxim from the 1960s and 1970s still applies: garbage in, garbage out. More broadly, AI adopters were encouraged to first properly understand – and then prioritise – the business problems they were seeking to address. Don’t start with the technology, start with you and your customers’ needs. To this end, one delegate borrowed from Abraham Maslow ‘law of instrument’ which states that: “If the only tool you have is a hammer, it is tempting to treat everything as if it were a nail.” In the context of AI, the delegate cautioned that “not everything is a nail. My advice: choose carefully.”
‘Powering AI’s potential’ – a Tech Monitor roundtable discussion in association with AMD – took place at the Hilton Frankfurt City Centre, Frankfurt on 23 April 2024
