Thought Leaders
“We're At the Top of the Crest of the AI Wave”
How SMEs are the next big growth area for artificial intelligence
As one of the key players in enterprise artificial intelligence, Hewlett Packard Enterprise is well placed to know what’s next. Lucy Ingham speaks to chief technologist Matt Armstrong-Barnes about how small and medium-sized enterprises are emerging as the next boom in artificial intelligence
In the business world, AI has been the hot topic of 2018. The opportunities for enterprise artificial intelligence have infected almost every industry with ardent excitement and technology has fallen over itself to keep up with demand.
One of the leaders in this space is Hewlett Packard Enterprise (HPE), which has been producing hardware in partnership with Nvidia.
“We're seeing huge leaps and bounds from HPE and Nvidia,” says Matt Armstrong-Barnes, chief technologist at HPE. “We've recently co-developed our Apollo 6500 range, which is specifically targeted at AI training workloads, and our strapline is AI from the cloud to the core.”
As a company firmly focusing on AI, HPE is witnessing the enthusiasm for the technology.
“The number of use cases that we see is enormous,” he says. “Between HPE and Nvidia we've got more than 4,000 partners that we work with, as well as building AI into our product sets.”
However, Armstrong-Barnes is also clear that AI is not a wonder technology that can take over any aspect of an organisation.
“AI today is a statistical method and what really it's good at is aggregating large amounts of information and providing those as a data pipeline into an existing business process,” he says.
“AI is not going to replace core banking processes or core processes; what it's really going to do is provide the right information to people.”
So far, however, its benefits have mainly been felt by larger companies. But with SMEs set to be the next key growth area, another AI boom is on the way.
Artificial intelligence for SMEs: the next big growth area
While many industries are embracing AI, with banking and financial services leading the way, the next big growth area is not, in fact, an industry vertical, but small and medium-sized enterprises (SMEs).
“I think the next evolutionary step is going to be getting into the small and medium-sized enterprise because at the minute, it's kind of the big boys: they've got the data, they've got the data scientists, they've got the understanding, and as a result, they can understand how they can put it into the business processes,” explains Armstrong-Barnes.
“Outside of some specific use cases, on the small and medium-sized enterprises front, really it's going to be the next evolutionary step.”
He argues that this is because the groundwork has already been undertaken by academia and large-scale industry, meaning the technology is now becoming more accessible for SMEs.
“We're at the top of the crest of the AI wave. The volume of data that is generated, label data that can be used for training, the algorithms we've got are significantly improving. They've been around for a long period of time, and the hardware you can run on is becoming much cheaper and much better at handling AI workloads,” he says.
“If you take that forwards, the next thing that's going to happen is it's going to grow out where it's much more accessible to smaller or medium-sized enterprises.”
This is something that HPE is beginning to witness, with innovative smaller companies seeking larger partners to embrace the technology.
“We are seeing early adoption from some of the small and medium-sized enterprises. They don't have the data scientists or the data to build their own AI models, so what they do is they'll partner with someone like HPE or Nvidia,” he says.
Such partners can provide pre-trained AI, as well as providing expertise about how to best apply AI to specific use cases. However, not all verticals are seeing the same enthusiasm from SMEs as yet.
“The main areas of growth we see are financial services, manufacturing is another big one, and then there's quite a lot of stuff we're doing in law. Areas where we see growth are pharmaceuticals, and healthcare – Nvidia have been doing some great stuff.”
“It's going to be the next evolutionary step.”
Fast, targeted and measured: getting started in enterprise artificial intelligence
Whether big or small, for businesses that are just getting into AI, there are some key lessons to learn.
“The first one is: think about the use cases. Everyone says 'I want to do everything with AI'. You need to get targeted. You need to think of minimum viable products, so think fail fast, fail quick, fail gracefully. If it doesn't work, move on to the next thing,” Armstrong-Barnes says.
“The second one is: get some data that's going to underpin that. And the third thing is put that data out there to be used by people.”
Conversely, there are also some key don’ts for applying AI to a business.
“Don't use AI to mask a bad business process,” he says, giving the example of using AI to sit over a paper-based system.
“AI doesn't read paper. Well it can, but if you're using it because you've got a document, someone is filling it in, you're digitising it and then using AI to read it back again, that's not effective use of AI, it's going to be expensive and it's not going to fulfil the business case.”
For companies, the central lesson is to pick the right target and make use of partners to effectively deliver on your goal.
“Make sure you're doing it with the right thing, make sure you've got the data to support it, get to your users early because they generate more information, more data,” he says.
“Make sure you've got the right platform on which it's going to run, and critically, use partners. Either build your own partner ecosystem or use somebody else's partner ecosystem.”
“Don't use AI to mask a bad business process.”
AI is never finished
While smaller businesses will often work with pre-trained AI, it is important to remember that there is no such thing as finished artificial intelligence.
“AI is a journey, not a destination. It is something that you continually need to do, work on use cases, once you've got some successes, continue to build on that,” says Armstrong-Barnes.
For companies, this means making use of the data generated by in-use AI and feeding it back to make improvements.
“AI only gets better when you do more training. So as you gather things that you infer in the field, take that back into your models, do more training and make them better,” he says.
“And by getting into that process, what you want to do is you then start to pick up new use cases, you start to embed AI thinking into what you're doing because it does require a little bit different thinking.
“One of the things is: where does AI fit in an organisation? Is it a business problem? Is it an IT problem? Is it a data problem? Is it something that users do? And actually, AI sits right in the middle.
“If you think about it, HPE, we're a technology company. So we can come up to you, we can find all the technology, we can do all of the technology work around it. What do we not have? Your data. Your understanding of what your business problem is and how your data relates to that business problem. So you need to put the two together.”
“AI only gets better when you do more training.”