Microsoft launches new product to overcome AI ‘paralysis’

Microsoft is launching a product to get artificial intelligence into the hands of more Australian workers as companies face a ‘paralysis’ of adopting the much-hyped technology at scale.

Microsoft Cloud and AI Group vice-president Scott Guthrie.

Microsoft is launching a new product to get artificial intelligence into the hands of more Australian workers as many companies face “paralysis” with adopting the much-hyped technology at scale.

Australian companies have been eager to explore AI but many are struggling to move their projects from trials to production. Tech advisory firm ADAPT found 77 per cent of Australian companies were “ineffective” at generating value and despite the technology accounting for the “lion’s share” of innovation budgets was failing to deliver on its promises.

This is creating a roadblock to the government’s bullish forecast that AI will inject up to $600bn a year into the national economy by the end of the decade.

The vice-president of Microsoft Cloud and AI Group, Scott Guthrie, said he understood the concerns many executives were feeling but was adamant the technology would help turbocharge productivity and make life easier.

After all, plenty of global companies are finding success in trials, with Microsoft’s AI business on track to surpass an annual revenue run rate of $US10bn ($15.49bn) next quarter, making it the fastest-growing division in the tech behemoth’s history.

“What we’re seeing is a lot of organisations are able to put together a proof of concept or a demo that has them very excited,” Mr Guthrie said.

“In other words they can show ‘gosh if we could make this scenario work, we could drive better customer support, or we could drive better insights into the data sets that we already have, or we could enable better self serve inside the organisation, around business processes’.

“And then the question is, you’re a bank, hospital, or you’re in any industry for that matter, how do you go from a demo concept, or a proof of concept?”

And this is where the hurdles start piling up and eating into a company’s risk appetite.

A Deloitte study found that 70 per cent of organisations had moved 30 per cent or fewer of their generative AI experiments into production.

ADAPT principal analyst Shane Hill said “a lack of good data is causing analysis paralysis at the executive level around this technology”.

“Organisations aren’t sure which opportunities to chase, which ultimately ends up with the only innovation Australian companies bet on is AI,” he said.

“This is problematic as simpler, less risky forms of innovation are being neglected for a technology we haven’t yet built the infrastructure to use safely.”

Mr Guthrie said the challenge was making sure that technology worked without sparking outages, inappropriate results or security breaches.

“Now I’m going to run this as a mission critical application for the bank, I need to make sure that every deployment works, and it works with high quality – so I’m not bringing down the system when I make updates,” he said.

“I need to make sure that if some employee decides to get creative and tries to have the AI say something inappropriate, that I have a safety system that prevents abuse.

“I need to make sure that I don’t have any weird biases in my solution, or anything that causes data skewage or weird hallucination results. If there’s any problem … can I get the right developers to engage to fix it?

“And that’s, I think, where a lot of these proof of concepts people kind of stare at and go, how do we do all that?”

Mr Guthrie said this was why Microsoft had launched Azure AI Foundry, which brings together its collection of AI models and tooling with new capabilities to help organisations customise, run and manage AI solutions, including custom agents and copilots at scale. And he said it could be done in “hours or days” instead of months.

“That really accelerates what AI solutions can go into production fast. And again, gives companies the confidence that they’re going to get a solution up and running quickly.

“That is going to be secure, it’s going to be reliable, it’s going to be safe, and ultimately, it’s going to give them the business value that they’re looking for.”

Mr Guthrie said he understood the hesitation many companies were feeling.

He said customers were using AI in ways that “weren’t possible two years ago”. This includes Pimco, one of the world’s biggest financial firms, creating its own chatbot to deliver “fast, relevant information to their teams”.

“A financial analyst or a trader or a salesperson can now, in natural language, figure out what product should I offer or trade to this particular client, and how do I answer their questions?”