I scroll down my Facebook feed and notice my cousin got married. But it was posted 3 days ago! Meanwhile, on Twitter, the tweet at the top is 20 minutes old, the one under that 3 hours old, and the third – the freshest – 9 minutes old. So, the social media goliaths have harnessed artificial intelligence to decide the sequence in which I want to read content. How infuriating!
Recent news headlines (Iltalehti 12 March 2018) report on a plan to establish a colony on Mars as a refuge from a third world war on Earth ignited by artificial intelligence (AI). Such media coverage only adds fuel to general frustration with the hype about AI. Most news seems to either extol AI’s epoch-making opportunities or to paint a gloomy picture of apocalyptic threats – often predicting massive job losses. Calming statements by experts in the field are buried under the hype and hysteria.
As became apparent at the outset of the Elements of AI online course, a generally accepted definition of artificial intelligence does not even exist. An essential component is the ability of a computer or program to function autonomously and to learn ‘by practising’. Applications are often associated with interpreting numbers, text, images and speech, or with producing them by machine. Researchers at Stanford, for instance, fed 62,000 cat photos through an algorithm, and got it to recognise in which photo there was a cat. Admittedly, MIT’s researchers later tricked Google’s image recognition algorithm into identifying a cat as guacamole.
There’s some way to go before a machine writes our emails for us, takes care of our children and elderly relatives, or even drives a car reliably in our winter conditions in Finland. Nevertheless, if recognising a cat succeeds today, you would think recognising a cancerous cell would be possible. In Finland alone, there are many companies, such as Aiforia and Mvision, developing machine-learning algorithms to facilitate cancer cell recognition. Finnish expertise is taken to the forefront also by Utopia Analytics, which uses AI for moderating online discussion platforms, and Optofidelity, whose robot uses machine vision to automate the testing of touch-sensitive displays. Other fields, also, encompass a wide range of highly specific but value-creating tasks that can realistically be automated.
This is not the first time that artificial intelligence has been at the peak of its hype cycle. Postulating high-flying visions has its place, but if expectations are not kept at a realistic level, we will experience the aftermath of previous hype cycles once again, when the reality gap quickly dampened the interest of companies, researchers and investors. That would be a pity. The corporate world is still full of unautomated manual work and still awash with data – in fields ranging from finance to health – not yet fully utilised. We need artificial intelligence so that we can ourselves concentrate on something more productive – or more fun. Like those cat photos.
- Investment manager in Tesi’s Venture Capital team.
- Education: Aalto university, Master’s degree, Engineering physics and mathematics, Systems analysis, Operations research.
- Former experience: Royal Bank of Scotland, Client Coverage, Helsinki, Nokia, Financial Analyst.