IT must learn how to manage the business, as well as systems, to secure investment in AI
Gaining support from the business to pursue investment and training is the biggest barrier to progressing with artificial intelligence, said panelists at Computing ‘s AI & Machine Learning Live event this week, with limited timescales putting on the pressure to come up with immediate fixes, rather than working on long-term solutions.
« 70 per cent of the activity that companies are doing is traditional business intelligence, » said Auto Trader’s David Hoyle, quoting Computing research, « but the real value from AI and machine learning will come from predictive capabilities. »
Auto Trader has a huge focus on website availability – understandably, as it is the portal through which it conducts all of its business. That can mean a trend towards short-termism:
« Sometimes it’s firefighting short-term problems: it means that you sometimes tend to focus on the BI aspects… It can be a challenge to draw the company away from that and say, ‘We’ve had a look at some longer-tail timescales here; these projects might not yield any benefit initially, but they will yield a benefit down the road’. »
Natalia Konstantinova, lead software engineer at First Utility, agreed: « You need to have an expectation that this process will take a bit longer, » she said. « It brings a lot of value, but it’s not going to work out of the box like magic…
« From a proof of concept you see how much value it brings, and then you try to push this small thing that already works… If [the business] see something small is working, it’s much easier to promote this technology and move further. »
Gaynor Collins Punter, IT director at the Health Research Authority, also thought that managing the business was important, but highlighted a different area:
« One of the biggest challenges is management of expectations and hype. People quite often think that things are further along than they are… To go from doing things in a very basic way, to doing things like [artificial intelligence], it’s management of expectations around all of the work that was discussed earlier – all of the data preparation, the standardisation, but most of all: what are the use cases and what is the business value? »
