A 2019 study by the National Standards Authority of Ireland found that almost 40 per cent of Irish companies currently use AI solutions and over 50 per cent plan to deploy it by 2024 across business, industrial and public sectors in Ireland. There are more than 126 AI and machine learning startups in Dublin alone.
AI as a technology can be split into different areas such as natural language processing (NLP) and computer vision (and much more). Currently, one of the more obvious uses of AI in today’s business is language models. These can be seen all over social media where questions are answered automatically. Often there is a prompt, like drafting a search on Google, and the AI offers options to complete your search.
Most of what we term current AI is better-termed machine learning or glorified pattern matching. To explain, linear regression is the means to predict the value of a variable based on the value of another variable. In this example, an X Y graph with three different points that line up in a rough line — by varying the slope and intercept, the weights if you will, you can eventually end up with a straight line.
Machine learning follows similar principles, but instead of three data points, there can be billions of data points and instead of two dimensions, there can be many more.
Again, the power of language models is that they are pattern matching, using many different kinds of sources to mimic human speech.
Language models use the vast data sets already on the internet where they learn to predict the next word and this also requires massive data centres. AYLIEN is an AI, NLP & Machine Learning startup based in Dublin, providing text analysis and news APIs that allow users to make sense of human-generated content at scale
In computer vision, humans provide labels in a supervised format. Many thousands of images of cats for example would have to be named as cats so that the AI can eventually come to distinguish a cat from a dog.
Examples of local Irish computer vision startups include Swan, using video and AI to make correct predictions when it comes to sizing and ordering clothes online, and MoveAhead, a motion-tracking platform designed to integrate movement and physical development into children’s increasingly digital worlds.
Richard Blythman is the founder of Video Force based on Trinity’s campus which works with sports training and recovery. Using motion capture systems where exercises such as the common squat are filmed from many angles in a controlled setting, the mapping learned by the machine is then applied to real videos of sportspeople taken on a smartphone.
“We use a technique called human pose estimation where we are looking to analyse the skeleton of a human just from a video. We can detect the movement in 2D and then we estimate what 3D might look like.”
However, we are only scratching the surface of what we can actually do with AI. Mark Yeeles, vice-president of Industrial Automation, Schneider Electric UK and Ireland, uses AI to fight climate change.
“IntenCity is the most efficient Schneider Electric flagship building, located in Grenoble. Equipped with a Microgrid Advisor, it can interface with other buildings in the neighbourhood as part of a local network, with the option of opting out in the event of high demand for electricity or a high tariff. In doing so, it can choose to store renewable energy produced onsite, and defer consumption in favour of neighbouring buildings.”
Blythman has subsequently founded Algovera to empower online communities of data scientists to own and monetise their work. This is important as the IP from most AI projects are held by the large tech giants or the universities that sponsor launchpads.
“Centralisation of IP means the control of this powerful technology is in the hands of a few making it difficult for machine learning engineers to work for themselves,” says Blythman.
“We are looking at using IP NFTs to facilitate ownership and control of assets created and sharing of any value generated.”
To draw an analogy with the way humans work, learning is one component of intelligence, but there are others like reasoning and higher functions of intelligence. This is where AI will move in the future, adding greater value than just predicting the next word or turning off a light.