Despite Tech Challenges, AI Is Not Experiencing a "Winter" in 2020

Artificial intelligence may not be living up to the hype, according to a compelling set of recent articles from The Economist.

The hype

PWC predicts AI will add $16 trillion to the global economy by 2030; McKinsey projects $13 trillion. AI proponents have predicted that the technology would be fully capable of taking over driving or medical diagnostics jobs by now.

The hype cycle

Tech nerds get hyped thinking about the Gartner Hype Cycle, which charts a technology's "it factor" over time. Emerging Tech Brew's Twitter account, for example, is located at “the slope of enlightenment.”

Leading up to the hype cycle: Researchers coined the term "artificial intelligence" at Dartmouth in the summer of 1956. In the decades that followed, AI faced two "winters" when funding and interest in the tech waned. AI’s “Innovation Trigger” came in the 2010s, when AI researchers made technical breakthroughs with machine learning and neural nets.

Now, AI may be sinking into the “Trough of Disillusionment.” Productivity is limited in 2020. For advanced research labs like OpenAI or DeepMind, training state-of-the-art models costs millions of dollars in computing power. As VC firm Andreessen Horowitz argued in February, AI companies face scaling challenges, shallower business “moats,” and smaller margins than traditional software businesses due to cloud computing usage and reliance on human handlers.

My takeaway

While AI faces real technical challenges and the rate of advancement may be slowing, we're not experiencing an AI winter. Larger companies and investors are still making deals (see Volkswagen's $2.6 billion investment in Argo AI earlier this month). And AI remains a top national security priority for the two largest economies, U.S. and China, as well as their largest tech companies.