23 September 2020, a rainy day in France, and I was peering through my computer screen into the AI-enabled Tech Foresight Summit, which I had organized together with Itonics, and Rohrbeck Heger and which I had planned to attend on-site in Berlin. But with our newly acquired status as ‘covid-19 high risk area’ I had to settle in among our 600 online participants.
Now a month later, I would like to share my three key insights in three blog posts:
1. If AI is the answer, what was the question?
2. Is Europe too late for AI?
3. AI for all?
Having graduated from a technical university, it does not take much to fire me up on promising new technologies. Artificial Intelligence (or simply AI) has made it into the vocabulary of every Fortune 500 CEO and probably features also on their top-buzz-word list to select from for shareholder meetings. In the latest Gartner Hype Cycle, 50% of all featured technologies are AI or AI-related technologies.
AI is believed to take over tasks across industries from manufacturing to financial services to medical diagnosis and treatment. In particular for binary answer to complex questions AI has taken a strong foothold. The success of the most successful FinTech company, Ant Group, is largely due to its ability to offer financial services and loans to clients that find it difficult to work with established banks and the basis of this is large amount of data from various sources that AI algorithms process to evaluate prospective clients and their requests. In financial services, AI is already at work on approving or declining credit requests, monitoring stock markets and making performance forecasts for companies after their initial public offerings.
Yet, our panelist Benno Blumoser, Innovation Head of Siemens AI Lab, adds a note of caution: “We need to have realistic expectation around AI, as it’s not a magic tool it is hard work”.
Claudia Pohlink, AI Head at the Deutsche Telekom Innovation Laboratories, explains that the winning combination is not just a set of data and a data scientist, we need good domain experts, the right questions and … patience.
Benno adds that transparency is also key to build trust in AI. We also need robust AI, which has safeguards against biases that have accidentally been trained-in, that for example discriminate against certain user groups and we need good safeguards against data breaches.
The biggest threat to the success of AI seems to be however the lack of AI literacy. The winning AI use case will require the domain expert, the top management and the regulators to be sufficiently AI-literate to make smart choices. In 2016, Jeff Immelt of General Electric who said that every new hire at GE will have to learn to code. Maybe in the future it will be AI literacy that we all will have to acquire…In the next post I will explore the question if Europe is already too late to the AI party.