Kathryn Hume, VP of product and strategy at Integrate.ai, offered some early perspective by saying that AI is not a new concept. In fact, the exploration of AI has been going on for well over 50 years. So what’s the difference now?
“We are swimming in data,” Hume said. “There’s a tremendous amount of it, and we are seeing an increase in the importance of the machine-learning algorithm. ... The thing that makes it all possible is the drop in parallel-processing costs.”
However, we’re still in the early days of a true AI revolution, with many questions yet to be answered around what the tech is and is not, according to Susan Etlinger, an industry analyst at Altimeter Group, who also moderated the panel.
AI’s impact in the enterprise today is twofold. The first is AI-powered analytics systems built using data and algorithms, and whose decision-making gets better over time. These systems enable organizations to collect, store, and process data at scale, which, in turn, helps them make informed decisions for the future, Hume said.
AI also is helping organizations become more efficient with its ability to “sense, think, and act to achieve a set of objectives,” said Anand Rao, partner and global AI and innovation lead at PwC. He pointed to natural language processing, the ability to identify and classify objects, reasoning, problem solving, planning, and even simulating as examples of capabilities.
“We as humanity are having this awakening,” added Chris Duffey, head of AI innovation and strategy at Adobe (CMO.com’s parent company). “There is this technology out there that accelerates and augments our thinking. Framing it in that sense, the sky is the limit.”
He also predicted that if an organization is not using AI in a strategic way by 2020, it will inherently lose market value.
Tom Goodwin, EVP and head of innovation at Zenith, said that he thinks about AI as a “transformative change” within the enterprise. He is most excited about AI’s opportunity to “do better things” from a customer experience standpoint.
The Future Of AI
The open-source movement, according to PwC’s Rao, is going to accelerate what can be done with AI. The big opportunity will be a better understanding of the customer and “bringing insights to bear throughout the customer life cycle,” said Catherine Havasi, chief strategy officer and co-founder of Luminoso.
Theodora Lau, founder of Unconventional Ventures, said she expects drastic improvements in the systems that help us predict what will happen in the future. After all, she said, there’s only going to be more data, which increases accuracy in the AI algorithms.
But it won’t be totally friction-free, and it’s likely that companies will spend just as much money to get the data out of their existing systems as they did to get the data in, said Robbie Allen, CEO of Infinia ML.
“It turns out to be the long pole in the tent,” he added.
PwC’s Rao said he is most excited about opportunities around simulation. He provided the example of a financial services company with a tremendous amount of data on its customers. AI can be used to analyze their financial profiles (salary, assets, debt, etc.) and make predictions about what their financial situations will be 20 years from now. Taking it one step further would be offering intelligent recommendations on what account holders can do to change their financial future.
“[This is] moving from segments to individual people,” Rao explained.
Infinia ML’s Allen expects AI to help organizations reduce costs and impact the workforce in terms of allowing people to be more productive and, as a result, achieve more breakthroughs. AI is going to be a “super power for humans,” he said.
Chris Benson, chief scientist of AI and ML at Honeywell, agreed. “So many people are worried about job loss, [but] for us it is about augmenting the workforce and arming them with the tools they need to get their jobs done,” he said.
In fact, given the financial risks, it would be unwise to not have a human there for checks and balances, PwC’s Rao said. After all, “people with computers almost always beat out just computers,” added Jana Eggers, CEO at Nara Logics.
Advice For Enterprises
Now is the time for organizations to experiment with AI, Honeywell’s Benson said. His advice: First assess the problem you need to solve and then figure out the technology to use.
“Having that assessment ensures you are picking the right tool,” he said, which is a big factor in “whether you are going to be successful.”
Companies need to start with small AI experiments--and expect failure. At the same time, they should be ready for a full-blown rollout should an experiment work, Integrate.ai’s Hume said.
Companies don’t need to run out and hire the top PhDs from the best schools, either, Nara Logics’ Eggers said. Instead, they must figure out what data they have and then just “get out there and try things, find experts, and learn.”
To minimize AI’s biases, PwC’s Rao said companies “will be better off” by building many AI models versus just one. The right approach, he said, is an “ensemble model,” which is similar to crowdsourcing in that you get multiple POVs.
One last piece of AI advice: “It’s better to walk than just start running,” Infinia ML’s Allen said. “Your data strategy must be in place first.” Then, figure out your AI strategy, he recommended.
Culture, talent, processes--the entire organization mentality--is all going to be different in a world where AI is in the enterprise, according to Unconventional Ventures’ Lau. “It’s all about the customer experience,” Adobe’s Duffey said. “Start there.”
Watch the entire panel discussion here: