LONDON – With the AI market anticipating 40 percent compound annual growth rate (CAGR) until 2024, the technology is moving well beyond the initial proof-of-concept (PoC) phase into real-world deployment and scaling. Enterprises are no longer merely embarking on AI experiments—they’re well into a journey.
The increase in the number of AI applications is undeniably impressive but that doesn’t always mean smooth-sailing. The field is plagued with talent shortages, hardware challenges, and crucially, data obstacles: how it is collected, curated, and stored. With these issues proving critical obstacles to effective scaling of AI applications, enterprises need a stronger understanding of what needs to be dealt with—and how soon. Nowhere is this more needed than in the crowded and highly competitive space of financial services.
Patrick Smith is Field CTO EMEA at Pure Storage, one of the foremost data and AI infrastructure vendors in the market. Prior to joining Pure, Patrick accrued 25 years of experience in the financial services industry. Today, he brings that experience to his role at Pure, where he oversees the use of technology across its large international customer base. Ahead of Pure Storage’s participation in The AI Summit London, we caught up with Patrick to get his insights into the opportunities and challenges ahead.
Q: Hi Patrick. What insights have you gained from your experience on both sides of the AI equation—as a customer and now as a vendor?
P: AI is one of today’s most interesting trends, and one that is growing incredibly quickly. Every industry is now not only thinking about whether they need AI—they’re one or two steps ahead of this. Most have started to realise that there are problems which, if they don’t solve using AI and emerging technologies, someone else will.
Q: What does the AI journey look like for enterprises today?
P: We’re starting to see organisations feeling the pressure to begin their AI journey. The public cloud is emerging as an ideal place to start that journey by using a rapid low, or no-commitment test to prove the business case for an AI workload.
What then happens, however, is that the public cloud workloads become prohibitively expensive. We’re seeing customers reach a tipping point where deploying on-premises infrastructure allows them to do some more of those experimental activities without there being a constant financial burden.
People start in the cloud, prove value, prove a business case, then look to move towards on-premises to really exploit all of the benefits from their data without incurring additional costs. Ultimately, it’s all about hitting business objectives and aligning AI strategy with what companies want to achieve.
Q: So what are some of the problems facing enterprises that AI can solve?
P: The problems are largely business problems. Part of solving that is making use of the data they have or gaining access to data, and that’s where you suddenly realise the scale of the challenge in terms of compute resource and datasets. Often, it’s a case of asking: what’s the business problem? Have I got the components necessary to solve it? How do I best align those components to deliver the business value as quickly and economically as possible?
Another big consideration today is delivering that value as accurately as possible. Accuracy is becoming an increasingly significant concern. A great example would be the recent developments in San Francisco around banning local agencies’ use of facial recognition. It highlights the challenge that industries face, firstly in terms of gaining confidence in the use of the technology, and secondly, that governance aspect around AI.
Related: There’s an AI arms race in financial services – could hybrid cloud storage be the silver bullet?
Q: When it comes to ownership of AI projects, where do you think the power lies?
P: What we’re finding in enterprises today is largely a coming together of business and technology. They’re increasingly working on the same page. There’s a realisation that to deliver quickly and easily, you need unified teams pulling in the same direction.
Lots of organisations are also creating AI-specific job functions. It’s not unusual to find a ‘Head of AI’ within an organisation. These people are joining the dots between business problems and technology solutions and effectively providing the glue in the middle between business challenges and the data necessary to solve those big problems.
Q: What are the key trends people should watch for in AI from now until 2020?
P: As good as AI solutions are today, there is still a lot of room for improvement and that will be driven by an ability to access more and more data. There aren’t really any shortcuts here. Being able to get data quickly and easily will be central to the success of AI initiatives well into the future, and data volumes are just going to keep growing. I don’t think that’s a trend we’ll see ending any time soon.
Patrick will be presenting on the use of AI in financial services at the AI Summit on Wednesday 12th June at 4pm. You can visit Pure Storage’s booth (AI200) on either day to learn more, or visit the event website here.