At first glance, NVIDIA isn’t a company that immediately comes to mind when we talk about artificial intelligence (AI) strategizing in enterprises, although it indeed plays a central role in AI because of its dominant position as a GPU (graphics processing unit) supplier. It is GPUs that enable the deployment of interactive graphics and of supercomputing capabilities on processors and at enterprise edges that facilitate the intense processing that AI applications require.
SEE: Hiring Kit: Computer Research Scientist (TechRepublic Premium)
Recently, I visited with Tony Paikeday, NVIDIA senior director of AI systems, about the strategies that CIOs are using to successfully introduce and integrate AI into the enterprise.
“The interest in AI has pivoted from hand-waving, science fiction-like applications to a growing number of pragmatic use cases in companies,” Paikeday said. “In part, this is being shaped by the COVID-19 environment that we are presently in. Companies are adjusting to a new normal and may have seen an inflection point where AI is a now strategic fit.”
Paikeday said that companies are focusing their AI strategies on three primary areas: Improving customer relations, streamlining business processes, and trying to position the investments in AI that they are making today for better results in the future. “All of these areas have well-proven use cases,” he said.
SEE: Robotic process automation: A cheat sheet (free PDF) (TechRepublic)
In company call centers, AI technologies like natural language processing have enabled agents to better understand customer sentiment along with the context of business transactions. The net result is that call center agents are more empathetic to customers. This builds customer loyalty and also assists companies in retaining customers.
“Being able to analyze the text of call center calls taps a gold mine of information that companies can use,” Paikeday said. “Without AI, it is estimated that companies capitalize on only 2% of this information.”
In sales support, AI can evaluate product inventories and then match inventory stock points to product demand. The net result is that less inventory has to be carried, which results in cost savings.
In manufacturing, work processes are evaluated to determine whether facility square footage can be reduced, and whether human and machine effort can be saved.
In industries such as oil and gas, field inspections are simplified and streamlined for cost savings, thanks to the assistance of drones and AI. “In one case, a company is saving $100 million a year in inspection costs, and over $500 million in repairs because maintenance can now be predicted before equipment fails,” Paikeday said. “The research suggests that by using a combination of AI and drones for field maintenance, companies can reduce their maintenance costs by 25%, their actual inspections by 25% and their maintenance downtimes by 15%.”
In food and beverage, Dominos Pizza has deployed AI to predict pizza delivery times. The AI analyzes how long customers wait for deliveries, and it assesses delivery wait times against the number of employees working at a given store, the number of orders in the pipeline, the complexity of orders, and the heaviness of traffic. “They have reached a point where predictability of when an order will get to a customer is now at 95%,” Paikeday said.
SEE: AI in 2020: How use cases will drive artificial intelligence deployments (TechRepublic)
3 strategic steps that enterprises can take to capitalize on their AI investments
1. Acknowledge the reality of shadow IT
Shadow IT, where end business units bring in technology without involvement from IT, is here to stay. Instead of IT trying to contain or control shadow IT, Paikeday advocates that companies establish an AI Center of Excellence that both IT and end users utilize. “This breaks down the silos between business units and IT, and gets everyone working together,” he said.
2. Standardize AI tools
As part of the AI Center of Excellence, companies should strive for standardization of the AI tools that IT and citizen data scientists in the business use.
3. Revisit your IT architecture
“As enterprises move to cloud, many are discovering that they’re running up their costs,” Paikeday said. “We are now seeing an inflection point where more enterprises are deciding to in-source their AI to their corporate data centers. This places the AI closer to the actual users who are using it. That being said, enterprises will continue to operate on a hybrid IT model that uses both internal data centers and the cloud. What they’re finding now is that they may need to rebalance their deployments between cloud and the data center to achieve best cost and performance.”