Published on the 04/04/2024 | Written by Heather Wright

From GitHub Copilot to Gemini Advanced, Anthropic and the AI Act…

Damon Kelly is open: Navigating AI options is tricky, especially given the rapid pace of change in the market.

But the founder and CEO of New Zealand software developer, system integrator and website designer, Enlighten Designs is very clear: “You need to start playing with it and understanding this topic – it is not going away, and it is valuable today.”

“It’s phenomenal. We find ourselves bouncing ideas off it more and more.”

His company, whose clients run the gamut of sectors including manufacturing, utilities, agriculture, education, travel and tourism and iwi, is already reaping benefits from its use of AI, and he’s keen to see other local businesses follow suit.

But he admits that not all companies are aware of the generative AI tools available, while others are ‘slightly dabbling in it but not in a very well thought-out way’.

Using Enlighten Design’s own maturity index, most companies it presented to scored themselves a level one for maturity.

A 2023 report from Datacom on the adoption of AI across Australian and New Zealand AI use within local organisations surveyed was high in Australia, at 72 percent, but New Zealand was far more subdued with 48 percent saying they used ‘some form’ of the technology. What form that usage takes wasn’t spelled out in the report.

More recently, a Forsyth Barr report this year showed modest adoption and investment in AI for Kiwi companies. The New Zealand Corporate AI Survey showed many companies remained ‘stuck in first gear’ with little change in their AI stance in the past two years, despite the generative AI onslaught.

“Some of the more advanced tools are cost prohibitive in New Zealand and a lot of companies are choosing not to use them,” Kelly told iStart.

There are, however, a number of tools which aren’t cost prohibitive and which Kelly says companies could, and should, be using.

“We are helping customers internationally build the tools into their products – particularly Azure Open AI – to enhance existing applications with better understanding of human input/language and also, we are finding new opportunities for products that didn’t exist prior to large language models having that level of human understanding.”

As a software development company, Kelly says GitHub Copilot is used ‘a lot’. That’s because it’s already built into the tools the company uses. Data privacy is also a key consideration and he says because it’s a Microsoft product, it is perceived to be more trustworthy.

“GitHub Copilot improves our productivity by leveraging AI to suggest code snippets and complete lines of code, enabling developers to write code faster and with fewer errors.”

He says that translates to faster development times, potentially higher code quality and the ability to keep up with the latest coding practices, providing clients with more efficient, lower cost and innovative software solutions.

But, while Enlighten has seen it make senior developers ‘far more productive’, Kelly is open that it doesn’t have the same effect on more junior developers.

“Our main uses of AI are GitHub Copilot,” Kelly says. “It’s phenomenal. We find ourselves bouncing ideas off it more and more.”

It’s being used mostly to prompt thinking, with some of the team using it in ‘many aspects’ of their roles now, especially for creativity-based jobs.

AI is being used to automate potential frameworks, including for content creation, and for business analysis assistance.

“It can be compelling articles or points but also for our clients – it’s great to help collect information and store it into custom GPT’s so that it can then be used as a repository for anyone to obtain information after we have completed a project.”

It can then be used as an assistant that can talk through a discussion Enlighten’s team have had with the client.

“It is definitely not a substitute for talking to the clients but sometimes we can’t remember all the things we have communicated about.”

The ability of AI to infer certain things that aren’t clear is also useful, he adds.

“If there are questions from developers rather than coming back to you, they now have a virtual representation of the client they can ask.”

GitHub isn’t, however, the only tool being harnessed by Enlighten. The company often uses large language models for the same queries, harnessing GPT, Gemini Advanced and Anthropic.

“We have found that Anthropic is a favourite for content creation, but we still heavily lean on GPT.”

The company is also finding work in helping customers modify their websites to improve discoverability and profile in a world where users are now turning to GPT or Copilot, rather than Google, to search.

For companies in Australia and New Zealand, as in the rest of the world, there’s a(nother) AI elephant in the room now: The EU’s AI Act.

Last month members of the European Parliament voted to pass the act, dubbed the world’s first comprehensive law on AI.

It will bring into force AI regulations, potentially as early as this month or next, which will see categorise AI based on its level of risk and prohibit some systems, while imposing stringent regulations on systems deemed to be high risk.

The Act applies not just to companies operating in the EU, but to any business operating there or handling data of EU citizens, necessitating adherence to its regulatory framework to maintain market access and competitiveness.

“The EU AI Act is going to impact us significantly here in New Zealand, we can’t pretend that it won’t,” Kelly says.

“We are going to have responsibilities that come in because of this. We will need to adhere to strict compliance requirements and perform rigorous risk assessments and think about how the technologies we use will be considered by the end users.

“It certainly introduces another level of compliance management on providers who are wanting to sell and deliver into the EU market and local companies who deliver into Europe will be watching with interest how the bellwether cases develop.”