If artificial intelligence takes your job, the Government could pay you an income purely for being human - that’s the future this economist expects as tech companies talk up AI in this week’s earnings. Video / NZ Herald
“Garbage in, garbage out” is one of the oldest cautionary sayings in IT - and it applies equally to one of the newest technologies, artificial intelligence.
Less than half of large NZ organisations were using AI, according to a study commissioned by Datacom last year.
Howare things looking now?
“The action is picking up. Broadly, we’re seeing a shift from experimenting with AI now to now thinking: What are use-cases for where I can get value? We’re seeing more organisations starting to do things like deploying tools like Copilot from Microsoft to look at productivity, or how use AI to augment customer service,” says Datacom NZ managing director, Justin Gray.
New research released by Datacom shows New Zealand businesses may not have the quality of data to capitalise on AI and all it offers.
The Wellington-based firm commissioned Cuia to survey 200 senior decision makers at organisations with 100 or more staff. From the results, it extrapolates that:
Just 9 per cent of New Zealand businesses consider 100 per cent data to be “clean” - that is complete, and free of inaccuracies, duplication, inconsistencies or outdated material.
Nearly half (47 per cent) of New Zealand businesses believe that “close to 75 per cent” of their data is clean.
30 per cent of companies think that half or less of their broad data set is free of issues.
“The quality of AI outputs is wholly dependent on the quality of data that is available – garbage in, garbage out,” Gray says.
“For example, AI can be a powerful predictive tool and can identify patterns and trends within your business that you can use to guide strategic decisions, but if the data is off, then the AI analytics will be too.
“There’s potentially a higher risk of tainted information.”
Good intentions, not so much action
The research found businesses have the right intentions for their data, basic hygiene practices around data management are lacking with 40 per cent admitting they do not conduct regular data cleansing and 13 per cent unsure.
Data audit and assurance protocols are also lacking with 46% of businesses stating they did not have these in place and a further 11% being unsure.
Time for a cleanse
“Data is the digital residue we leave behind when we digitise everything, and there’s lots of it, and we’re keeping it because it’s relatively cheap to store,” Gray says.
Where do you begin, if you decide it’s time for a cleanup?
“It can seem complex and daunting,” Gray says. “Sometimes people don’t know where to start because they can’t quantify it. But I think really what it comes down to is starting with looking at how you’re storing it, how you’re classifying it.”
Datacom N Z managing director Justin Gray: Risk of an AI delivering tainted results if it's working with "dirty" data.
“It’s about treating your data as a product and making sure it’s got the right ownership for someone who thinks about how it needs to be treated as a product or an asset that needs to be maintained and then giving them the mandate to start looking at,” Gray says.
Focus on key areas, and accept that you’re not going to get to 100 per cent.
There are technical elements to a cleanup, such as identifying and dealing with duplicates, but there are also legal factors such as data sovereignty, or making sure files that need to be stored in a New Zealand data centre are indeed onshore.
When asked about specific barriers or issues being encountered when utilising data, the surveyed businesses identified the top issues as
Incomplete data: 51 per cent.
Duplicate data 46 per cent.
Inaccurate data 43 per cent.
Low-quality data 39 per cent.
Compliance or regulation issues: 34 per cent.
Privacy risks: 30 per cent.
Security risks: 22 per cent.
The research also indicated some were more optimistic about the security of their data than their policies and systems warranted. It found:
91 per cent of New Zealand businesses believe their organisation has adequate data and security protocols to protect their information.
But only 60 per cent of businesses have data governance policies and procedures in place, managed by trained staff.
Data integrity is not just about keeping it safe from hackers. Again, governance issues come into play - which again feeds into one of Gray’s central themes.
“Classification is one of the most important areas to start because we’re seeing a lot of organisations that haven’t done the classification. When you start pointing artificial intelligence at your data, if it doesn’t know what information should be served up to who, you start to have risk of privacy issues.”
Chris Keall is an Auckland-based member of the Herald’s business team. He joined the Herald in 2018 and is the technology editor and a senior business writer.
