Generative artificial intelligence (AI) has created a significant buzz in the past two years. In doing so, it has arguably overshadowed more mature AI technologies that already have a proven track record of delivering business value.
Preparing for AI, however, can be complex. Organisations need to not only identify business cases where the technology will improve things, but also ensure that the data they have is capable of producing results.
“I don’t really think that anyone is 100 per cent ready, perhaps with a few exceptions,” said JC Durbin, head of AI innovation at technology consultant and software developer Ardanis.
The problem is that while a great many organisations are keen on becoming data-driven, which is to say making rapid decisions based on real-world information, their data is noisy, typically unstructured and, more often than not, siloed.
A 50 to 200-person SME can do a lot [with data and AI] by working with a company like ours
Breathless claims about AI have raised expectations, Durbin said, but the reality is that data still needs to be put in order. After all, as the old computing adage has it: garbage in, garbage out.
“There is a sense that AI is a magic box into which you throw your unstructured data and then get the results you want. The truth is, it doesn’t work that way,” he said.
Instead, a thorough data cleansing process is required, as is an understanding of which systems data is actually stored in.
Durbin said it was possible, given the breakneck pace of development in AI, that this could change. Howerver, the reality was that, for the foreseeable future, producing quality results from machine learning (ML), data analytics and AI required a close focus on the data itself.
“Do I think that will change? Eventually. We may well have AI that can be used to make sense of data that looks very incoherent to humans, but that is not what it is capable of doing right now. AI does move very quickly, though,” he said.
SME intelligence
Nevertheless, organisations that do develop AI, ML and analytics techniques are seeing tangible benefits to the bottom line.
Interestingly, while much of the debate about how to make the most of data has focused on large enterprises and the public sector, smaller outfits can benefit too. In fact, in many cases it is easier for them to do so, as smaller businesses have smaller datasets. Indeed, Ardanis works both with enterprise and SME-level organisations.
“A company that consists of three people may not be able to benefit much, but a 50 to 200-person SME can do a lot by working with a company like ours,” Durbin said.
Practical tasks include summarising information to assist with sales and customer support.
“In a situation like this it’s not about trying to understand 30 terabytes of data. Perhaps they sell insulation material and have 600 datasheets that need to be scanned and summarised.”
This is in marked contrast to large enterprises where needs can be extremely complex. But if the workload is large, so too are the potential rewards.
“The enterprise is where it is really, really hard,” said Durbin, “and this is why people like Microsoft are putting in huge amounts of work to make it possible.”