TECH & AI
What challenges do you foresee in implementing AI-driven solutions within the energy sector , especially when it comes to infrastructure upgrades and regulatory compliance ?
Sophie : In heavily regulated industries , a common challenge is that they tend to be more set in their ways of working because the regulator mandates it and they get used to working in those ways .
I think it ’ s about behaviour change , being more agile and receptive to different potential use cases of AI and also testing the value AI is bringing .
At IFS , we have a real focus on the time-to-value of AI implementation . How long does it take to start seeing business benefits ? There ’ s a lot of hype around things like generative AI , but in many cases , it has yet to prove its benefit . We have a real focus on what that time -to-value looks like so we can help the organisation get comfortable with the investment in AI .
Although we ’ re now talking a lot about generative AI , some of this industrial AI that the IFS is in has been around for a while . This is not something completely new . We have tried and tested use cases , customer success stories and can model out what that looks like in terms of the financial benefits , the productivity benefits and the sustainability benefits . I think for me that ’ s where we have some really good tried-and-tested AI use cases and that helps people get comfortable .
Kevin : There are a couple of things that are important to note here . One is that everything hinges on the quality of data . This means we might have mountains of data that we can go through and check for anomalies or guidance , but if that data is not good , the outcome won ’ t be good . The first thing has to be a quality set of data with a true strategy around what outcomes we hope to get from that .
When I look at even IFS ’ acquisition of Copperleaf , this gives us the ability to also layer on top of what would be the best approach for our customers in terms of ranking those investments .
There are a lot of things to look at , but data is the key piece here because if we have incorrect data , we can have hallucinations and therefore the wrong recommendations .
The second important thing is that AI offers a lot in terms of automation . It gives us the ability to get rid of some busy work and perform deeper analysis , but if we ’ re removing human components of that , where are the checks and balances to make sure that what we ’ re constantly modelling and tweaking is what we expect and what we want to come out of the system ? We must have the availability of checkpoints in that automation to make sure that AI is functioning the way we designed it .
energydigital . com 65