Artificial intelligence (AI) is proving to be a brilliant product of human innovation. It automates a wide variety of business processes to the scale of the internet, of email, and of digital media. It can do things that humans never could — like predict hospital outcomes from Electronic Medical Records and explore the galaxy using datasets too massive for humans to fathom.
While a significant innovation in itself, AI only produces optimizations for existing practice. It detects patterns, but does not comprehend the physical world. AI can inform and inspire, but does not produce innovation per se. Algorithms that automatically recommend media, place advertisements and detect fraud are all built upon explicit assumptions —that the training data is sufficient, and that the state of things will remain the same.
Of course, things never remain the same. Markets change, patient care evolves, science and technology move forward. With each progressive stage of an organization, AI needs to be re-trained and re-purposed, much as a factory needs to be re-tooled to produce a different product.
For example, AI trained to autonomously design a new office chair isn’t able to conceive of the ergonomic advantages of the standing desk. It doesn’t know what a chair really is — all it knows is the 3D shape of chairs as defined by its training data. To tie-in the David Foster Wallace cartoon above, AI lives in water, yet it doesn’t know what water is.
AI still has an important role to play in innovation, however. At my startup Tag.bio, we spend a lot of time investigating how AI and data analysis can support humans and accelerate the process. Recently we came across this Harvard Business Review study of 3.5 million employees outlining the most critical drivers of innovation within organizations. Here are the key points:
- “Scale — more participants.”
- “Frequency — more ideas.”
- “Engagement — more people evaluating ideas.”
- “Diversity — more kinds of people contributing.”
It just so happens that those drivers of innovation perfectly align with the core vision of our software platform — to unburden expensive and scarce data scientists, and utilize tailored User Experience (UX) — to make anyone a data scientist. Based on the feedback we’ve received from our initial customers (to be announced soon), I’d say we’re well on our way to that end.