Nobody disputes that 2018 will be a big year for AI, but as usual, the reality is significantly behind the hype.
Figures out from Gartner in late February show the gap between talking and doing. The 2018 CIO Agenda Survey showed that while 4% of organizations have implemented AI, another 46% are thinking of doing so.
This growth curve is natural for any emerging technology. We are at the beginning of the Industry 4.0 project. Business cases need proving. The technology itself is at an early point.
It takes a brave organization to be an early adopter. For many, it makes more sense to be a follower and watch what others do -- not so much to replicate their success but to avoid their mistakes.
AI-savvy Australasian Biz
The Gartner global statistics offer a helicopter view of the state of AI deployment. Down here in the Antipodes, Australian finance solutions provider Maia Financial is doing more research.
Maia Financial operates an Equipment Demand Index (EDI) in both Australia and New Zealand, polling businesses with revenues of between AUD 5 million and AUD 250 million a year. The polls ask them about, among other things, the nature of the assets they are planning to acquire.
The two indices are reported separately, but both show that Australasian businesses are more advanced in their deployment of AI than the overall Gartner results.
In Australia, for example, the growth over the last year has been significant. In December 2016, only 1.6% of the 500 businesses surveyed said they planned to acquire AI assets in the next quarter. By October 2017, this figure increased to 5.2%.
Over in New Zealand, the percentage moved from 1.5% to 3.3% in the same period, which is more in line with the Gartner results. However, we need to bear in mind that the Maia Financial results indicate which assets business will purchase in the next three months.
Industries Adopting AI
Cumulatively, the Australasian research suggested that if all the organizations which said that they are planning to implement AI followed through, deployment should be between 5-10% across the board.
Maia Financial results also tell us which industries in Australia are more engaged with AI. These tally with the McKinsey “heat map” results that show healthcare, financial services, manufacturing, and mining are the ones that are investing in AI so far.
We don’t see the Australasian hospitality sector getting into AI so far, and adoption in agriculture is weak (they are focussing on drones, but AI could come later when they start crunching the data they have collected).
Already, Gartner has some lessons for any fast followers gleaned from the experience of the early adopters, and these speak to the early-stage deployment of the technology.
It makes sense, as Gartner noted, to aim low at first. Think small, prove the technology and get some rewards from pilot projects, and then think about next steps.
Taking Out Human Bias
The Gartner research also shows that as humans, we are woefully unprepared for the advent of AI.
Fifty-three percent of organizations rated their ability to mine and exploit data as "limited" and Gartner estimated that 85% of AI projects in the five years through to 2022 would deliver erroneous outcomes due to bias in data, algorithms and the teams responsible for implementing them.
It is clear that the big issue in deploying AI is not AI itself, but humans. It is why Gartner suggested that success will come from augmenting humans, and not replacing them.
Take humans out of the equation, and that is where the dystopian fantasies can run amok.
Living with AI
With the Internet of Things, the nightmare is of hackers infiltrating and controlling billions of devices, with AI the fear is not just of losing your job but living alongside AI’s which have their existence and agenda beyond serving humanity.
In Australia, Professor Genevieve Bell articulates this issue. She was originally an anthropologist who had a two-decade career at Intel in the US, where her role was to humanize the firm’s cutting-edge technology.
Now heading up the new Autonomy, Agency and Assurance Institute at the Australian National University in Canberra, Australia, Prof. Bell spoke publicly about the need for the right governance around AI.
The danger, as she saw it, is in developing and implementing such a transformative technology without understanding the big picture social impacts.
Ultimately, AI will be inside so many of the daily applications, services, and objects we use that most people in their everyday lives will hardly notice it. Just as we never think about the workings of an internal combustion engine when we start a car.
Gartner’s survey speaks to organizations who are using AI as part of what they do but is still relevant to Prof. Bell’s concerns.
Having people engage with AI and use it as part of their job, by dedicating workers to "monitoring and guiding" neural networks, and through choosing "transparent" AI solutions we can build trust and AI can contribute to human utopia, and not a dystopia.
It is something that the 46% of CIOs looking to implement AI should bear in mind, Gartner noted.