Ai is here, and foremost amongst the concerns of workers in manufacturing is replacement. The question is whether that fear is justified. MEPCA’s editor attended the recent Westminster Employment Forum: The Future of AI in Employment to learn more about this divisive topic.
“Ethical adoption requires clear governance, human oversight, transparency, especially when the investment is starting, [and] trust building with employees.” Nina Gryf, Senior Policy Manager, Make UK.
AI adoption has become synonymous with progress in the manufacturing sector, and beyond. Whether used in the form of the latest ERPs, intelligent CAD programs, or tools for predictive maintenance, AI-based technology is already proving its usefulness.
However, there currently exists a hyper-focus on AI in the media, treating it simultaneously as a solve-all solution and dangerously out of control technology. But is AI the seismic shift in industry that we’re being led to believe? And are the fears of widespread job replacement valid? Or, in both cases, are we overreacting and overreaching based on the information available?
It is with these questions in mind that I attended the Westminster Employment Forum policy conference: The future for AI in employment – priorities for policy, regulation and industry best practice.
The opening speaker, Professor Joanna Bryson, Professor of Ethics and Technology, Hertie School, Berlin, began the discussion by pointing out when we speak about robots or AI taking jobs we’re prescribing agency to them that they simply do not have; it is better to say that corporations will make decisions on how many people they hire. Jobs are, she pointed out, agreements between people.
How technology affects an industry, Bryson suggested, is not always as we might expect, using Geoff Linton, one of the founders of machine learning and Nobel Prize winner, as an example. In 2016, he asserted that people should stop training radiographers, in reference to the disruptive power of machine learning tools introduced in radiography. In reality, the demand for radiologists increased, with more clinics being able to house a radiographer.
Of course, it does not follow that technological advances mean more jobs, necessarily, but Bryson’s point is that it can do. The analogy Bryson used: software that doubled the efficacy of teachers could result in schools that were twice as good or half as many teachers (or a combination of the two). The difference, Bryson pointed out, “is a matter of policy. It’s normative.”
Reframing this in manufacturing and it’s the difference between a business being able to expand its operations, or reduce headcount and achieve the same output. AI based tools could enable both; the choice is the manufacturer’s, unless policy mandates, prescriptively, the more ethical choice.
During the course of the conference, two illustrative comparisons stood out. The first, which was introduced by Professor Joanna Bryson and later echoed by several other speakers, was that of the industrial revolution. By no means a novel comparison given AI’s prominent role in ushering in industry 4.0., but it serves to show the importance being placed upon the technology.
The second comparison of note was first introduced by the chair Viscount Camrose, Shadow Minister, Department for Science, Innovation and Technology, when he likened the hype-cycle of AI with the dot-come bubble, which saw the unprecedented growth of investment in internet-related technologies before bursting in March 2000, resulting in a dramatic market crash. Camrose’s fear is that AI, as the current focus of trillions in investment, will lead to a similar crash. This is something we will return to later.
Before the second panel discussion, Trinh Tu, Managing Director of Public Affairs at Ipsos UK, shared research into public opinion of AI, which addressed important factors around AI adoption, including the impact on the job market, and the skills gap. Showing in particular that a lack of trust is proving to be a real barrier to adoption: 55% of people did not believe that AI would create more jobs than those that it replaced, with only 12% believing it would.
Trinh Tu: “And these anxieties really translate directly into the public’s distrust of AI’s benefits.”
Interestingly, Anglosphere countries, including the UK, are more nervous about AI adoption than other global demographics.
However, it did not follow that the UK is anti-innovation, as Tu explained, but rather it is concerned about the slow reaction of policy in comparison to AI development, with 60% believing that “UK government should adopt a cautious AI strategy to safeguard jobs and allow adaptation, even at the cost of global innovation pace.”
Tu concluded that to improve trust and therefore further AI adoption organisations should clearly explain the use cases for AI, including what task it is for, and why it is of benefit to that task, etc.
Following on from this broader view of AI adoption, Nina Gryf, Senior Policy Manager, Make UK, provided an industrial view point. Gryf acknowledged that the UK has several great research institutions developing innovative technologies, but referring to the report Future Factories Powered by AI[1], Make UK surveyed 150 manufacturing companies and found that only 36% were using AI in their factories.
This comes despite the fact that over half of the companies using AI are already seeing improvements in efficiency, productivity and reducing operational costs. The reason for this reluctance, echoing the wider research undertaken by Ipsos UK, is one of confidence in the safe use of AI.
Gryf illustrated how industry might combat this using the example of one of Make UK’s members who involved workers in the purchasing process of AI-powered robotics, enabling them “to be part of the decision making”, which proved to help adoption and alleviated fears of job-loss, reinforcing the notion that the workers were being upskilled rather than replaced.
Gryf: “Ethical adoption requires clear governance, human oversight, transparency, especially when the investment is starting, [and] trust building with employees.”
In addition to these points, which are crucial to building trust in AI amongst workers, the wider application of AI and views around return on investment are also important. As mentioned above, the current hype around AI has drawn comparisons to the dot-com crash of 2000.
John Chadfield, National Officer, Technology, Communication Workers Union, added weight to this comparison, making reference to both Klarna’s disastrous AI deployment and Meta potentially reducing its AI department.
“We’re also far enough into the AI hype-cycle that longitudinal research is now available on a number of impact factors, and this includes MI TS report published last month, as stated 95% of organisations found zero returns despite enterprise investment.”
Chadfield went on to address some myths around AI before ending, controversially, on the notion that the main thing holding AI back is neither adoption or data, but simply that it doesn’t work, and furthermore, what enterprises appear to be trying to solve with AI is the ‘problem’ of paying wages.
In the following discussions, Nina Gryf, commented that in manufacturing, there isn’t evidence of layoffs, as those described by Chadfield, as a result of the adoption AI and automation; instead the sector has a very different problem: over 50,000 vacancies that need filling, and a growing skills shortage.
In manufacturing, at least, investment in AI is less likely to be driven by the ‘hype-cycle’ of Wall Street investment, and more by practical concerns.
In the last panel, Dominic Lusardi, Non-Executive Director, Digital Thinkers, made a sobering point: that we should “demystify” AI; “It isn’t magic,” and he described it simply as a system of “pattern recognition running at speed over a zero latency network. That speed is extremely impressive, but it’s also a risk. Once an error is in the system, the A I will repeat it endlessly until a human steps in.”
While the long term benefit of AI and how it will impact industry and the job market remains uncertain – and divisive – what is certain is that it will lead to change, and how to manage that change and ensure it is utilised positively is with policy, transparency, and the involvement of those it will affect most.
westminsterforumprojects.co.uk
[1] makeuk.org/docs/future-factories-powered-aipdf/download?attachment
