In a matter of days, Microsoft will unveil the much-heralded new version of its Copilot software to a business world already severely disappointed by Big Tech’s initial AI offerings. It also comes hard on the heels of a stern warning from Gartner to organizations across all sectors that the cost of introducing artificial intelligence (AI) to the workplace could easily balloon by a staggering 500 -1,000 percent.
But Microsoft’s current marketing push for its latest AI offering, a souped up version of its Copilot service, is rapidly gathering momentum, in spite of commercial AI’s dismal performance to date. Microsoft chief executive Satya Nadella is currently touring 39 cities around the world with new products and use cases for AI. He predicts that the performance of AI systems will double approximately every six months, and the AI revolution is about to be led by a souped-up version of the company’s existing Copilot software, part of the 365 package.
“The question now is how do we transfer this to the real world…Think of Copilot as a user interface for AI,” Nadella told an audience in Berlin.
But Nadella was unable to resist drifting into the kind of Silicon Valley over-hype that has so far falsely raised expectations of the nascent new AI technology. He attempted to impress his Berlin audience with a rehashing of the old science fiction fantasy of “having computers that understand us instead of us having to understand computers”. According to Nadella, this is now a reality, conjuring up sci-fi images of supercomputers more powerful than that of any human who ever lived, such as Hal 9000 in the 1960 Stanley Kubrik movie 2001: A Space Odyssey.
Reality of AI is a little more prosaic
But the reality of Microsoft’s offering on its current world tour is a little more prosaic. The software giant is, for example, showing a use case involving German automation company Thyssen Krupp Automation Engineering, a major supplier to Germany’s automotive industry. Microsoft has worked with Siemens to develop a bespoke software, Siemens Industrial Copilot, to help with the setup and maintenance of complex machines. The in-house co-pilot for industrial applications can, for example, support the programming of the system by translating instructions into correct automation code. Microsoft says Honeywell has also seen productivity gains equivalent to adding 187 full-time employees from the use of Copilot, which raises the big question hanging over the tech giant’s new AI offering. Will it enable an organization to grow while reducing staffing levels, or will it prove to be a costly and over-hyped hairball?
According to Colton de Vos, Marketing and Communications Specialist at Resolute Technology Solutions: “I don’t think that Copilot will replace human staff next month because it is not a 1:1 comparison and replacement. Various tasks and aspects of a job become easier with more intuitive and useful tools – but it’s never as clean cut as a direct replacement. As with any new tool or way of working, processes and policies need to be built around using it so that it can scale properly within an organization. This all takes time.”
So far, however, cybersecurity is the only sector with real-life experience of a truly effective use of AI – committing cybercrime and extortion on an industrial scale while creating scurrilous and obscene but highly convincing video deepfakes.
Pavlo Tkhir is the chief technical officer of Euristiq, a digital transformation company that donates 100 percent of its profits to the Ukrainian war effort in the ongoing conflict with Russia. He is sceptical of claims that AI, particularly in the form of Microsoft’s hastily rushed-out Copilot software, will start replacing huge swathes of the global workforce anytime soon.
AI may initially be more staff-intensive
“It’s true that Copilot can be very effective in automating manual tasks. It’s especially great that it eliminates human error when it comes to technical repetitive tasks. But I want to stress that AI is, after all, only an assistant and cannot be considered a replacement for human workers,” says Tkhir.
He adds that, in it is early stages, at least, AI may end up being more staff intensive as it currently needs constant supervision in order to ensure it’s working effectively and it needs to be constantly monitored for bugs, biases, and updates. In fact, Tkhir thinks that new roles might emerge because people with relevant expertise will be needed to manage AI.
“On a more practical level, such overreliance on AI can simply disrupt your company’s operations. So, even if businesses start rapidly adopting AI and letting staff go, it will have its consequences, which will most likely require them to do damage control,” warns Tkhir.
I don’t think that Copilot will replace human staff next month because it is not a 1:1 comparison and replacement. Various tasks and aspects of job become easier with more intuitive and useful tools – but it’s never as clean cut as a direct replacement. As with any new tool or way of working, processes and policies need to be built around using it so that it can scale properly within an organization. This all takes time.
