AI is at that awkward age. While digital leaders pile in, many companies have only tipped a toe in the water. It's fair to say that the founder and CEO of H2O.ai, Sri Satish Ambati, wants that to change and his mission is to democratise AI's power; oddly enough, Covid-19 may accelerate the process.
One odd effect of the pandemic has been to make wannabe data scientists of us all, desperate for numbers to inspect and analyse. That hasn't been all good, clearly, and Ambati laments some of the false conclusions that have plagued the news cycle.
"There has been a pandemic of great proportions but also an epidemic of dashboards and a lot of misleading signals," he says, sporting his characteristic cap over a Zoom call. "We got better and better but in the early days there was clearly a lot of cognitive dissonance from poorly-made data science."
However, Ambati says that the pandemic may also play a part in speeding up acceptance of AI. He was already "expecting 2022 to be the year of true enterprise AI" but a tipping point could come sooner "because people have to bring in elements of holistic renaissance thinking".
Makers gonna make
A nine-year-old Silicon Valley company, H2O offers platforms and technologies that are in turn intended to spark a new generation of AI makers. He is simultaneously a cheerleader for AI but also a voice of reason: "All models are wrong, some models are useful," he suggests, and he knows that much of the power of AI lies in the benefits of making wrong assumptions over and over again to eventually home in on truth.
Ambati's vision is to see AI-infused data science everywhere and H2O is doing its best to help with tools and frameworks that are used by about 20,000 companies and about 250,000 data scientists. And he really does mean everywhere, even my own arcane and inky trade, where, he says, soon journalists will be able to test hypotheses with "a few clicks to produce stories that can be debunked or validated … facts or fiction."
Change and having the willingness of the senior leadership team to aggressively pursue new ways of working is a critical obstacle, however.
"AI is not a technology problem … it's a cultural problem," he says, arguing that more companies need to follow the examples of the Amazons and Netflixes of this world where continuous feedback loops based on numerous data sources whir constantly. Companies need courage but the "courage to act on data is a scarce resource, especially in companies that have 100 years of success".
Next steps
What's needed for the next stage of AI becoming pervasive is making it "faster, cheaper and easier". He says his company is all about the question of "how do you make it easy enough to do that customers can go and change the world". That also explains why he embraced open source: the race is on to become de facto and those companies that can get tools into as many hands as possible will be the likely leaders.
But speed is of the essence and it is at the heart of what H2O is trying to achieve. That is, turning science into software, schooling AI to do what humans do well and building app stores and recipes for processes and vertical sectors that act as fast-tracks to AI success.
"Our punchline for years and years has been ‘time is the only non-renewable resource'," Ambati says. "Speed matters because if you can come to a conclusion fast, then even if it is wrong, you learn from it and can go to the next level."
And, spurred on by cheap memory, CPU, bandwidth and cloud, speed is very accessible these days, making regression testing a rapid-fire phenomenon.
He believes that the rise of AI will mimic the sudden shifts in market leadership and company growth that the rise of the internet saw.
"Software has been eating the world and now software is the world but now AI is eating software," Ambati says, riffing on Marc Andreessen's famous dictum. "Software, hardware… every stack is being transformed by AI."
Of course, the pandemic has rocked and shocked business but Ambati remains a bull, predicting a strong fourth quarter for the US, bolstered by the digital economy and the bouncing back of segments such as transportation. He says we're still in the "early innings" for AI but adds that H2O revenues are already doubling year on year. With $147m pocketed in funding, he spies a future role as a public company, so long as the company sticks to its mantra of innovation and expansion. Building blocks are in place with Ambati saying the company has invested heavily in hiring data scientists and with a partner ecosystem of heavyweights including IBM, ATOS, Snowflake and the cloud hyper-scalers.
The H2O name has associations of transparency and neutrality, but he cautions that water is necessary for life but can be dangerous if not treated with care. The same responsibility applies to AI but, with the right guardrails in place, we are likely to see a new wave of businesses and ways of doing business.