Who’s jobs are safe and who’s aren’t from the robot invasion? It’s a hot topic right now (though arguably premature) and quite an interesting one from a business perspective writes DreamWalk founder and creative director Joe Russell
At this stage the mainstream conversation is very passive and dwells on the inevitable negative fate of the human workforce.
Rather than focus on the looming problem, let’s look at the solutions and potential opportunities. How can you and your business either beat your artificially intelligent competitors or use AI to your advantage in a post-AI landscape?
How to define AI
Firstly, let’s agree on a definition of AI. There is a lot of misunderstanding around what AI actually is and how far it has advanced. While IBM’s Deep Blue might be good enough at chess to beat world champion Garry Kasparov, its intelligence is limited only to the game of chess. Likewise, DeepMind’s AlphaGo is great at playing the strategy game, Go, but as admitted by its developers, if the size of the game board was changed even just a tiny bit, it would fail dismally.
While these kinds of machines are really great at throwing lots of computational power at a very specific task to achieve a result by brute force, this is not really like human intelligence. Merriam Webster defines intelligence as “the ability to learn or understand or to deal with new or trying situations”.
For the purposes of this article though, I’m going to include all levels of artificial intelligence and all applications of machine learning in my definition. Multiskilled and truly intelligent machines may pose the real threat to humanity, as warned by Elon Musk and Stephen Hawking, but it’s the basic, single-task workhorses that will be most effective at replacing humans in the short term.
What AI does best
So, the first step in figuring out how to beat or work with AI is to identify its strengths. Like Deep Blue and AlphaGo, today’s machines are great at accomplishing specific given tasks. The narrower the scope of its task, the better the Machine is at successfully completing it.
This doesn’t mean the task needs to be a simple one. Artificial intelligence is incredibly good at complex tasks like facial recognition, voice analysis, product recommendations and predicting human behaviour. The key is giving the computer a clear and specific goal, then training it with a sufficient amount of data to be able to achieve the goal.
AI is NOT currently very good at anything that requires a high level of creativity or creative thinking, a broad range of interdependent skills, anything that requires subtle or nuanced human behavioural analysis or situations where a limited amount of training data is available.
Pointing out the many subtle challenges facing complex automated driving systems, MIT Professor of Robotics, Rodney Brooks asks “Since there are no current ways that driverless cars can give social signals to people, beyond inching forward to indicate that they want to go, how will they indicate to a person that they have seen them and it’s safe to cross in front of the car at a stop sign?”
As is the case for the AlphaGo computer, deviation from a narrow and clearly predefined path of thinking is extremely difficult for the AI of today. This will probably be the case into the foreseeable future.
How to beat AI
Beating AI in business will become more and more a game of playing to your business’s human strengths and focusing less and less on the menial tasks that can be easily automated. This will likely require your business to pivot at some point or at the very least, reposition its service offerings to be more focused on its human capital.
A media buying company, for example, might be forced to reposition their services from ‘finding the most cost-effective ways to reach a client’s target audience’, which could easily be done by AI, to ‘personally negotiating the lowest media prices on behalf of clients.’
A transport company that chooses not to run automated fleets might need to focus less on the popular capital-to-capital routes, which will be dominated by fast and reliable self-driving trucks and instead specialise in reaching the more difficult regions where automated vehicles struggle (or are not allowed) and where human driving skills are required.
How to work with AI
Working with AI is becoming easier and easier every day. While the multi-skilled humanoid overlords of the apocalypse are a long way off, AI software that performs common tasks is relatively accessible right now.
Thanks to TensorFlow, Core ML and various open source advancements, mobile app developers can build smarts into applications easily and cost-effectively. You can even find a variety of pre-trained machine learning models online, which can be plugged into software applications to give them ‘intelligence’. These models perform a variety of common tasks like detecting sentiment in text, identifying objects in photos or translating spoken words into written text.
Goals and results
As the technical capabilities of AI improve every day, which aspects of your business you choose to automate will depend less on the technology available and more on your specific goals.
An online clothing retailer may identify up-selling as a new focus and choose to add intelligence to their product recommendations. While once the site would have simply suggested products other users that bought the same item had bought, AI could be used to learn about user behaviours and suggest more relevant products, therefore increasing up-sell conversions.
A fast food ordering app could identify speeding up the credit card entry process as a goal and incorporate credit card scanning technology, which reads credit card information from a photo of the user’s credit card so they don’t have to enter it manually.