Artificial Intelligence: What It Is And What It Is Not

Written by Holtby Turner

Degrees of separation between human and machine are inching closer. Human brains power robotic limbs; tens of thousands of workers are managed by artificial intelligence or AI, and chatbots act as digital replicas of us. “Humanity is, in effect, getting an upgrade” as Quartz observes in Machines With Brains.

Artificial Intelligence (AI) is not as new as you may be forgiven for thinking it is. With roots back to Alan Turing in the 1940’s, most of the deep learning we talk about today stems from neural network research in the 1980’s and 90’s – getting mainstream attention in 1997, when IBM’s Deep Blue beat chess champion Garry Kasparov.

Whilst a huge and highly diversified field, there is actually no such a thing as “an artificial intelligence.” Rather, as Associate Professor at NYU Julian Togelius says, it is “a collection of methods and ideas for building software that can do some of the things that humans can do with their brains”.

To understand which parts of an AI product or solution are automated, take out the specific sector or expertise, and then try to work out how just the technology could work on a different problem. This is why the technologies involved in AI especially offer such tremendous scale – they are not tied to any one particular vertical. AI driven solutions are horizontal and as such, can be adapted to other verticals. And that’s why we see similarities and cross over between PropTech and FinTech etc.

Automation doesn’t replace people. Automation increases efficiencies that bring down some labour costs so that productivity scales across and into others. Mortgage application platform Habito, is a fantastic example of AI working hand in hand with very human customer care associates.

Even where a certain task may be easy to automate manually from a technical viewpoint, to do so safely, consistently and to a high standard, is where the challenge lies when these tasks sit in unpredictable environments. A great example is in farming. John Deere, after more than 20 years trying to build self-driving tractors, learned a hard truth: there is no such thing as an empty field. They are dangers and obstacles everywhere. What challenges autonomous driving for John Deere and self-driving carmakers in general, is actually dust and other weather conditions. These impact changes in sensory reactions that humans intuitively use, and cognitive technologies are many years from safely replicating.

Translate this to ConTech and bricklaying or operating a crane – each has a repetitive, ordered process suited to automation – in an ideal, obstacle free world. But self-driving cranes on a high density, urban construction site? Would you want to be the first to take this to the market?

For real estate leaders and CEOs it’s rarely a lack of desire that holds back digital disruption and innovation – it’s the very human feeling of fear, scepticism and rejection – in all forms– that are undoubtedly the greatest challenge in leading change.

What does this all mean in practice? Well it means there’s no almighty “robot” coming to take our jobs; robots will not be building houses, or decorating offices, or managing projects in the foreseeable future. Just because something can be automated, will it be?

There’s the cost of developing and deploying both the hardware and the software for automation. On top of that is the cost of the labour. When the supply and cost of workers are cheap and plentiful, and the cost of automation expensive and complex, you’ll have a major adoption issue. Economies of scale are reached only when software and hardware reach mass adoption.

Robots replacing humans in manual labour roles such as cleaning, road sweeping, and fast food restaurant service will be a reality, but it will be done in stages and with very niche focuses.

Hyped headlines like ‘THE ROBOTS ARE COMING TO TAKE OUR JOBS!’ are just that – hyped headlines.

When robotics company Boston Dynamics brought human sized robots who could walk and climb to the market, they focussed entirely on strength and form. They terrified market observers with their now infamous designs, and went on to be sold to Softbank by Alphabet Inc. in a $500 million dollar acquisition.

Many in the AI and robotics industry believe it’s the cottage industry of bot-makers who seem set to trump the market by focusing on adoption first, and functionality second.

Concerned with what the machines look like, how they sound, and the kind of personalities they have, Mayfield Robotics in the US have hired engineers from Pixar to make their robots friendlier. As they say on their pre-order website HeyKuri, “for all the things Kuri can do, there are plenty of things Kuri won’t do. And that’s what makes Kuri a great robot.”

Recommended articles