The AI market is evolving very quickly, in two quite different directions.
The first, aligned to the IoT trend, is the application of AI to a given specific, focused use case – such as IBM’s Watson giving tax advice, self-driving cars, or consumer-focused online chatbots.
There are lots of these specific use cases, and generally they can or do operate in isolation: you don’t need any of the others for it to work.
The second flavour of AI is broader and more all-pervasive – the application of what might be called general intelligence to our day-to-day lives.
This may start with individual use cases, but aims to integrate them into a single AI – at the simplest level, a portal like Siri or Alexa which must be programmed for each specific request; and in aspiration something closer to the Computer of Star Trek.
Google’s progressive integration of its products and services with machine learning, or Apple’s tight-knit control of both software and hardware with predictive analytics (especially its ability to leverage the phone and the watch given the personal/physical data they can access), are the best examples of this latter aspiration.
But even the Swedish furniture giant IKEA is getting in on the action, launching a survey on AI through its Space10 division.
Apple, Google… and IKEA?
IKEA’s initiative seems superficially like a radical combination of the irrelevant and the grandiose.
But the group is uniquely well-placed to consider the ways in which consumers may wish to behave in their day-to-day environments – in response to AI as much as in response to anything else.
After all, IKEA has sold over 60 million units of its Billy bookcase alone – one for every 100 people in the world.
It has shaped our day-to-day environments, and it’s not daft for it to consider both the ways we may interact with technology in those environments, how those environments influence our interactions with technology.
Riding the wave or wiping out?
AI is an industry in which the big questions remain unanswered.
AI is an industry in which the big questions remain unanswered
IKEA’s survey asks, should AI be “autonomous and challenging” or “obedient and assisting?”
“Should your AI fulfil your needs before you ask?” or “should your AI prevent you from making mistakes?”
In the context of such enormous questions, the challenge for companies investing in AI is to navigate the twin risks of obsolescence and ROI.
Obsolescence risk is common to all rapidly-evolving markets. Will a competitive product leapfrog my own and render it obsolete?
obsolescence risk is common to all rapidly-evolving markets
While the risks of failing to generate adequate ROI apply to all investment decisions, not just AI or technology more broadly.
Companies can seek to mitigate these by focusing on specific market needs which are close to their current core activities. The more specific, well-defined and expensive the problem, the more likely it is that a given solution will find a lucrative market.
Similarly, the need for proper business planning is not unique to AI investment, but is particularly acute in a market that is evolving so quickly.
The role of M&A
M&A can help bypass some of these risks, as a company can acquire an established solution with a proven market position, rather than trying to develop and build from scratch.
good ideas will continue to blossom in smaller organisations and in universities
Startups and smaller companies are often better able to pursue blue-sky R&D without managerial pressure for immediate payback.
However, they often lack the financial resources to see projects through and may struggle to access the computing power and volume of data normally required for work in AI and machine learning.
A lot of good ideas will continue to blossom in smaller organisations and in universities, then to be spun out, raise venture capital to scale and/or to be acquired by larger incumbent players.
Incrementalism vs. integration
The areas of focus for these M&A targets likely to be in two main areas.
Firstly, complicated niches with direct monetisation potential and/or cost saving opportunities through streamlining and automating existing processes.
technology cycles tend to alternate between the specific and the general
Secondly, blue-sky transformational projects which re-define our understanding of the product or the problem, such as self-driving cars or all-pervasive general-intelligence products – or how to integrate AI into a sofa.
Technology cycles tend to alternate between the specific and the general.
Self-driving cars provide a good example – starting with the sci-fi vision of the 1960s, then evolution through various specific capabilities such as adaptive cruise control, keeping the car in-lane and self-parking, through to full autonomous operation.
The short to medium-term trend in AI will be incrementalism, as the market continues to pursue specific, relatively low-risk deployments for well-defined use-cases, before the pendulum swings back to integration.
the short to medium-term trend in AI will be incrementalism
The next stage will be to integrate a multitude of individual solutions into something than can solve larger, more complicated and open-ended problems – going far beyond the (admittedly increasing) number of pre-programmed questions one can ask Siri or Alexa.
AI will come full circle, as it moves beyond its original promise as a general or universal intelligence, through specific applications, and back in to an integrated capability able to address multiple needs without individual programming for each one.