The world is in the midst of an energy crisis, with Russia’s invasion of Ukraine the latest factor to drive up the price of gas and oil.
The UK government, for one, has stepped in with measures to help households bear the burden of price rises, although some analysts feel these did not go far enough.
Identifying ways to save energy is not always a straightforward task, especially for businesses – which is where London-based AI optimisation company TurinTech may be of help.
“Today the data regarding energy costs are public, competition in the space is intense with many different energy providers, and information about energy commodities is easily accessible,” CEO Dr Leslie Kanthan tells BusinessCloud.
“By putting this data together and applying AI, we can see the trends of how the pricing would change based on different scenarios that may take place. For instance, if gas prices go up, oil prices come up. Or, if there’s a shortage of water that could affect the need to heat less water, this would be more costly overall.
“Based on this, by using AI technology, it’s possible to have monthly or even quarterly forecasts. By analysing these forecasts, AI can help prepare yourself and your business to make energy-saving decisions.”
His company applies AI to optimise such models efficiently and profitably. Its evoML platform automates the entire process of creating, deploying and optimising models within days.
“When people design and develop models, there are a lot of inefficiencies and inaccuracies which may prevent them from going into production,” he explains. “Our focus is on finding ways to make those models accurate enough that they can go into production as well as ensuring they can deliver tangible cost savings to businesses.
“Some of these models can cost tens of thousands of dollars to compute; many businesses don’t necessarily have this kind of budget just to run on a specific model. Our role is therefore to build these models and optimise them within a certain cost budget frame, helping businesses save both time and money.”
According to Gartner, businesses are still firmly interested in AI, with 48% of CIOs having already deployed or planning to deploy AI and machine learning technologies over the next year.
Innovations in this space are happening at speed and being embedded into our daily life – including ways to use AI to control our energy consumption and predict our energy bills and ensure a more sustainable business.
“Right now, data centres consume huge amounts of energy: in 2020 global data centres’ electricity used accounted for around 1% of global final electricity demand,” says Dr Kanthan.
“A significant portion of these data centres are GPUs for computer-intensive operations. Being able to use code optimisation to optimise those algorithms can help get to the results faster and therefore use less energy.
“If you think that typically models can cost tens of thousands of dollars depending on what you are aiming to do if they are fully code optimised, these models can do the same thing for much less.”
TurinTech was founded in 2018 by Dr Kanthan, Dr Michail Basio (CTO), Dr Fan Wu (CSO) and Dr Lingbo Li (COO) after meeting at University College London during their PhD research. “Having all worked for well-known financial institutions, such as Credit Suisse, we were frustrated by the manual machine learning developing process and manual code optimising process: we found it to be time-consuming, resource-intensive, and require deep domain expertise,” says Dr Kanthan.
“Code optimisation is a missed opportunity businesses are not taking enough advantage of. The ability to optimise code from multiple different objectives, such as reducing the energy, the memory, the CPU consumption is a missed AI opportunity.
“Not only that: businesses can also use it to cleverly boost the speed and reduce the carbon footprint of their technology devices. A lot of hardware televisions, washing machines, dryers now use smart algorithms.
“With code optimisation, these devices can deliver quicker results using less resources to elevate customer experience and gain competitiveness.”
The company is aiming to attract more top AI researchers and academics to be part of its team.
“An important characteristic of code optimisation is that in a repository of six million lines of code it’s going to be humanly difficult for somebody to go in there and modify code improvement to make it more efficient and not heavily bloated,” he says.
“Today, we have a way to do that, and we will have even better ways of doing that tomorrow.
“The good news is that AI is still a long way from reaching its full potential. However, businesses need to look beyond the hype if they want to be able to benefit from its applications.”