Busting the ‘Myths’ of AI and Machine Learning
AI and machine learning are “part of your kit, but they’re not your silver bullet”, according to Merkle Aquila’s senior data analyst, Marta Portugal.
Discussions on the impact of artificial intelligence (AI) are often awash with hype, hyperbole and, on occasions, scaremongering.
“AI is coming to take all of our jobs” is a commonly-spun statement, underlining both a lack of understanding of the technology itself, as well as the inherent human fear of ‘being replaced’. Industries across the globe are being transformed by technology, with AI beginning to play an increasingly crucial role in sectors such as manufacturing, cybersecurity and the energy sector.
Speaking yesterday (1st of May) at DIGIT’s 6th annual Digital Energy conference in Aberdeen, the heartland of Scotland’s oil and gas sector, Merkle Aquila’s senior data analyst Marta Portugal and Analytical Consultant Dervla Brennan embarked on a ‘myth-busting’ crusade; underlining the benefits of AI and emphasising the need for understanding of this emerging technology.
“So, something that is in the press quite a lot is that the ‘robots are coming with their AI and they’re going to take all of our jobs’,” said Brennan. “Let’s be honest, they’re probably going to take some jobs, but mainly some of the manual jobs can be automated.”
Brennan noted: “The way we see it is that AI and machine learning (ML) are there to aid humans, as opposed to replacing them. What we need to be really thinking about is what are the new jobs that are going to be created when AI comes in and automates some of these more manual tasks.”
She added that, moving forward, the discussion should be centred around skills for the future workforce, and what will be required to carry out these new roles. This is a key issue that organisations, businesses and, likely governments, should be planning ahead for if AI & ML is to be incorporated successfully.
Breaking the Hype Cycle
Continuing, Portugal outlined what she describes as the “AI and ML Hype Cycle”, which detailed the many stages through which organisations pass when deploying – or considering the deployment of – AI and ML.
Over the past several years, Portugal asserted, she has witnessed a host of organisations traversing this cycle, with many currently still locked in it. Many find themselves starting with the ‘technology trigger’, as she describes it, which is essentially when companies have access to great technology or computational power and assuming it will solve all their woes.
Technological developments over the past 10 to 15 years has led many organisations to act impulsively, seek a quick fix and expend resources where they necessarily weren’t required.
Having access to this, she explained, “kind of led to everyone thinking that AI was going to solve all of their business problems,” – which despite throwing AI at every single problem, failed to offer any real solution.
“Everyone was very happy and had a lot of expectations (several years ago), but it obviously didn’t work out as they imagined,” she said. “People kind of lost faith in AI as a mechanism, or as a tool, to solve business problems, which leads them into a ‘trough of disappointment’.”
Portugal said that many businesses have fallen into this trap of assuming AI and ML is the key to solving their business problems or transforming and streamlining their operations. Simply put, without an appropriate strategy – or adequate understanding of the technology – they are not going to help solve issues.
“They are part of your kit, but they’re not your silver bullet,” she said.
Breaking the Bank? Think Again.
Brennan suggested that, as with many areas of business, significant investment is not always required; organisations should be smart and utilise a broad range of resources when introducing AI and ML into processes.
“It’s the same with everything. If you want to spend a lot of money on it you can, but you don’t have to do that,” she explained. “There is a lot of rentable architecture out there that is reasonably priced, so you can access this really easily.”
She added: “If we’re thinking in agile terms and thinking of a minimum viable product, for example, this can be built on a small data infrastructure, it doesn’t have to be built on massive data infrastructure.”
Ultimately, this is due to the fact that during this building process, companies will alter and add as they build, which will require flexibility.
“You can start small and build from there. It’s also likely that, as you go along, you’re going to be iterating and adding in bits and pieces as you build, so you need to be flexible,” she said.