FILE PHOTO: AI (Artificial Intelligence) letters are placed on computer motherboard in this illustration taken, June 23, 2023. REUTERS/Dado Ruvic/Illustration/File Photo
IT is like a new relationship – the first year was starry-eyed and full of promise.
Generative artificial intelligence (gen AI) was going to do your board presentations, write this newspaper column, delight customers and let every lay worker code away his own work problems.
But past a year of flirting and learning, bosses are discovering the extravagance and shortcomings – both the technology’s and theirs – in bringing home the hottest tech debutant since the Internet four decades ago.
Despite leading the region in government planning, findings are that just over one in 10 businesses in Singapore are ready.
Cisco’s inaugural AI Readiness Index compiled around September reported only 13% of firms here are prepared to deploy and leverage AI in general.“Considerable gaps exist across other key business pillars like infrastructure, data, governance, talent and culture,” said its president for Asean Tay Bee Kheng.
With ease of use arising from its response to common language commands, research firm Forrester expects gen AI to raise workers’ productivity by 50%, boosting them in areas such as self-service analytics, code writing and content creation.
But will it transform businesses? For only 30% of firms in the Asia-Pacific, said the firm’s senior research director Frederic Giron.
Even then, that will take place over years, not months, he added.
That gen AI is the biggest thing after the Internet is beyond dispute.
Bloomberg Intelligence projects the gen AI market to grow from US$40bil 2022 to US$1.3 trillion by 2032.
Its share of total information technology spending on hardware, software services, ad spending and gaming market spending will boom from less than 1% to 10% over the same period, the report noted.
Yet to get bosses to fully commit? That is going to take time.
Bensen Koh, consultant with tech advisory firm Access Partnership, expects firms to focus on areas with the “widest potential appeal, lowest perceived risk, and lowest need for human resources upskilling”.
For example, it could be used to help manage customer relationships and sales, accounting and finance, and for internal collaboration.
“These are functions that most companies have and would like to improve on,” he said.
“As they are internal facing, risk is perceived to be lower. So long as the AI tools are relatively simple to use, not much upskilling is required.”
Gavin Barfield, Salesforce’s chief technology officer for solutions in the region, said: “As many Asean organisations still have untapped data, it is projected that more companies will lay the groundwork for AI by 2024.”
Clear policies
Just getting companies AI-ready requires data governance frameworks, clear policies for data collection, storage and use, and constant monitoring and auditing of training data, he added.
For most back-office workers in human resources, finance and supply chain work, much more needs to happen before gen AI could support real-time analytics, flagging bottlenecks and inventory shortages.
Enosix, an integration provider for enterprise software giant SAP, blogged: Resources are stretched, data silos are killing productivity (and the customer experience), and AI is stuck in the starting gates without trusted data.
“But those aren’t SAP trends, that’s the state of the industry.”
When ChatGPT went mainstream in November 2022, it immediately seduced leaders’ imagination.
Announcements followed almost daily from businesses on new gen AI capabilities for customers and employees.
Some were embellished. Most were not meaningful.
Then, bosses became alarmed when they found workers privately engaging ChatGPT and handing over company secrets like source code.
Bot workers
They also began to realise the business risks of employing these bot workers that not only make things up, but also spew racist, sexist and other eyebrow-raising remarks.
At its 2024 tech trends presentation on Dec 5, Forrester senior analyst Liu Meng offered a teaser to his audience, predicting that in the coming year, “at least one insurer will offer an AI hallucination-risks policy. And it will be profitable”.
After July, when enterprise gen AI-embedded software such as Salesforce’s Einstein GPT, Adobe’s Firefly and Microsoft’s Copilot made pricing plans public, another concern arose: costs.
The risks and staggering investment sums ploughed into their partners, such as OpenAI and Cohere, that are building and running the large language models underlying the technology have been cited.
The joke about town is that these costs are about to give bosses another shock treatment, not unlike the one they got not long ago from cloud subscription charges.
Microsoft Copilot, for instance, sets a firm back US$30 per user per month.
But to even be able to use it, the worker must already be on the E3 or E5 versions of Microsoft 365, priced at US$36 and US$57 respectively, per user per month.
Commitment must be for at least a year.
So what shall one do if technology gets overtaken six months from now? And after shelling out for the technology, will workers put it to good use?
Forrester’s Giron suggests that firms start with pilots.
“There are ways for you to test and try with gen AI pilots which are low risk and low cost to go. And then you can double down,” he said.
“These things (like Copilot) are now going to become massively commoditised, embedded in software applications, which means that the case for justifying US$30 per user per month drops,” he added. — The Straits Times/ANN
Krist Boo is a senior correspondent with The Straits Times Singapore. The views expressed are the writer’s own.
