“The world has pivoted into AI in a big way.”
Ben Grubert isn’t wrong there: it seems that every company is suddenly keen to talk about artificial intelligence. Amid the popularity of generative AIs such as ChatGPT – capable of writing impressive copy and code in response to simple commands – harnessing the power of automation is very much on the agenda.
It is vindication of Grubert’s decision to found AI consultancy INEVITABLE with Sean O’Mahoney. The Manchester business aims to make innovative technologies such as machine learning accessible to all, working with startups and larger enterprises to develop bleeding-edge ethical AI.
There are levels of expertise within data science, says Grubert.
“A lot of people who approach machine learning will take data that they might understand, but they won’t really interact with it a huge amount,” he explains at INEVITABLE’s base in Bonded Warehouse. “Instead, they will use algorithms that they also don’t really understand.
“They simply know that ‘this tool will solve this problem’. If they have a whole bunch of people who listen to music, they can use a tool like matrix factorisation to recommend music to a new person, based on what they’ve previously listened to; but they won’t look into the mechanisms inside that machine learning process. They just become good at using the tools.
“Without that working knowledge, you become very good at applying it, but not improving it or building upon it or researching it.”
The human brain
Understanding the small moving parts – described as ‘lessons’ by Grubert – allows for the contextualisation of the information being processed, which in turn allows you to learn from it.
“If it takes 100,000 pictures of chairs to build an image recognition system that can recognise one, why does it then get confused when it sees a beanbag?” he asks. “And how can a child of three years learn to identify the beanbag more accurately from, essentially, a data set of one?
“It’s about how the brain works. In Alan Turing’s phraseology, humans essentially have better algorithms, and are able to contextualise the information based on previous experiences. So, as a result, we don’t need as much data to extract information.
“We intrinsically understand that a chair is something that you sit on. And we don’t have to build hard rules around that with multiple edge cases, because we don’t think about things like that – we think about things in terms of these features with blurred edges.
“The question ‘is a chair a chair, if you turn it upside down and you can’t sit on it?’ is something that we consider to be relevant. Different algorithms require different amounts of data to achieve the same effect.”
INEVITABLE’s services include development, CTO-as-a-service and technical due diligence for investors. Its core expertise lies in providing strategic guidance and technical solutions to help startups, SMEs and CICs to scale.
Grubert says it has worked with hundreds of startups – often providing free guidance and mentoring – since it was founded in 2019. “By understanding the moving parts underneath the hood of algorithms, we learn how to rearrange them according to very peculiar use cases,” Grubert explains of the firm’s ‘toolkit’.
“We are a mechanism so founders can have AI as early as possible, key it into their product. We can start to assemble the supercar from the ground up – here is the Maserati engine, the chassis, the wheels…
“The trick is to make sure it doesn’t just become a body kit.”
The company is also involved in ‘Tech for Good’ initiatives. It built COMPILEDMCR, which brings the Manchester tech scene’s events together in one place, making participation more accessible for the community.
Grubert will discuss the rise of AI, the evolving regulatory landscape and offer practical tips for harnessing ethical AI into your business in BusinessCloud’s Demystifying Tech podcast webinar. You can sign up at the link above and submit a question for Ben below
A Beautiful Mind
When Grubert first spoke with O’Mahoney at a tech meetup – “it was retro gaming evening… I was playing Street Fighter II” – they discovered a shared ability around algorithms.
“I came from a heavy maths background, while Sean had a computer science background,” he recalls.
“[After we began to work together] the walls of the incubator we joined were covered with maths, from floor to ceiling! I wasn’t worried that someone would nick our work – because you don’t leave maths on the wall if it’s resolved!”
I reference A Beautiful Mind, the film about a mathematics prodigy – played by Russell Crowe – which won multiple Oscars. Its protagonist would go on to win a Nobel Prize. “John Nash is actually one of my heroes, because he discovered game theory mathematics, and also did a lot in cryptography,” says Grubert.
Grubert and O’Mahoney’s company has a clear focus on helping early companies to establish ethical AI and data practices from the outset.
“I love sitting down with founders and figuring out where they are; where they want to be; and working out how to get there. By working backwards, we can say ‘actually, this is the data you want to be creating – and these are the ways that you can ethically collect that data’,” Grubert says.
“Most companies who are building something afresh, in our experience, only ever make unethical decisions when they didn’t know how to do it properly in the first place.
“It’s usually from a point of ignorance and ‘selling selling selling’, then thinking about it later, where they find themselves boxed into a corner where they then have to do something unethical.
“It shouldn’t be an afterthought. It should be a core principle. If it’s that important in the beginning, build it out of the box.”