Who’s doing your thinking? A brief look at Artificial Intelligence.
Have you ever had the feeling that your computer has a mind of its own? Or that your smartphone is listening to what you say? Are our devices really becoming cleverer than us?
The answer could be ‘yes’ to all those questions, in the future if not now. A fascinating exhibition at The Barbican called AI: More Than Human (16th May - 26th August) posits the idea of a world where human brainpower is not the only form of intelligence.
The Barbican exhibition examines the evolution of our desire to create non-human intelligence, which is an ancient one. It’s an intriguing if rather baffling journey through machine learning, deep learning, datasets, artificial life and embodied robotics. These terms may be incomprehensible to most of us, but the fact is that artificial intelligence (AI) is already embedded in our everyday lives.
AI is based on the concept of machine learning. Rather than trying to build computers that mimic the incredible complexities of the human brain - as scientists had previously been trying (and failing) to do, machine learning puts in place the fundamental networks that allow computers to learn like babies do; by repetition and gradually developing neural pathways, or algorithms, to make sense of words, pictures, emotions etc.
As the exhibition shows, AI’s uses have both benign and sinister implications but, in the words of former Apple, Google and Microsoft executive Kai Fu Lee, AI has moved from ‘rocket science to the mainstream’.
To return to smartphones, voice controlled intelligent assistants (Siri and friends) gather huge amounts of data about our activities and habits, particularly if we use voice search. There aren’t, as far as we know, vast secret rooms full of people listening in, but companies are certainly using the algorithms generated by our devices to record and predict our behaviour.
The marketing industry has already found many uses for AI. Organisations such as Amazon, Netflix and Spotify use it to record our preferences and make suggestions as to what we might like to buy next. It all comes from AI -based systems which continually adapt to our likes and dislikes.
AI has vastly improved the effectiveness of search engines. Google was the first to engage with it in 2015 when it introduced RankBrain, a machine learning based algorithm. Hence that feeling that your computer already knows what you’re looking for even when you’ve only typed in a word or two of your search.
Visual search, although still in its infancy, searches by looking for results that are visually similar, and is becoming more commonplace on platforms such as Pinterest and Google Lens. It’s applied in e-commerce by recommending products that ‘you may also like’, based on previous searches and purchases. In social media, marketers can use visual search to identify brands and logos to ascertain how and where consumers are interacting with their products and brands.
In the world of financial services, AI can be used to make decisions on personal loans in seconds by analysing masses of data on loans taken out by people in similar situations to calculate how likely the applicant is to default. Not only does this massively speed up the process - which benefits both the loan applicant and the finance company - but AI is already proving more accurate than humans at assessing risk.
Social listening and sentiment analysis may sound a bit Orwellian, but brands are finding it useful to discover what consumers are saying about them on social platforms. They can then use this AI to target people with relevant marketing messages.
Dynamic pricing is another product of machine learning, which analyses our data patterns to predict what we are prepared to pay, and what level of discount might entice us to buy. AirBnB uses this to advise property owners what rates to ask for.
AI has also made it possible to predict purchasing trends more accurately. Previously, it was only possible to guess at future behaviour based on data gathered from past behaviour. Now, marketers have a crystal ball at their fingertips in the form of predictive analytics, which can “reverse-engineer” our experiences and actions to determine which marketing strategies will be most effective.
Chatbots are a form of conversational AI and is something that many companies are planning to invest heavily in: by 2021, 50% of them will spend more on chatbots than app development. Trainline has recently developed a voice app that continually improves itself through machine learning. So the more it’s used, the better it will get.
Speech recognition, computer vision and augmented reality are further products of AI, all of which are working their way into our lives.
Should we worry? There have been a few reports from users of Alexa who found their device had recorded and forwarded private conversations.
Some of these occurrences do have a rational explanation however. The other day at the office, I was talking to a colleague about one of our property clients. Only seconds later, an ad for a development from this client popped up on my screen. As I wasn’t personally interested in the development, it did seem as though my PC had been eavesdropping. But it turned out that my colleague had been researching it himself. The advertiser had picked up the office i.p. address and AI concluded that I might be interested too.
There are of course legal and ethical challenges posed by AI, notably around data privacy and anonymity. And speaking personally, I am a little disturbed to learn that there is even an AI copywriting tool, developed by e-commerce platform Alibaba. It can spew out 20,000 lines of copy a second - but only in Chinese. Perhaps I don’t need to think about retiring just yet...
For a more intelligent – and 100% human – approach to marketing, get in touch - we’d love to hear your views.