Executives want generative AI, but are taking it slow. An army of providers has lined up to help

Enterprise companies are moving slowly and deliberately to adopt generative AI, if they have even started at all — whether because of concerns around enterprise data security and AI “hallucinations” or a lack of the necessary technology, talent and governance to implement generative AI successfully.

There’s certainly no doubt that executives want to access the power of generative AI, as tools such as ChatGPT continue to spark the public imagination. But according to a KPMG study of U.S. executives out this week, a solid majority (60%) of respondents said that while they expect generative AI to have enormous long-term impact, they are still a year or two away from implementing their first solution.

Companies can’t wait too long, said Martin Kon, president and COO of Toronto-based Cohere, which offers enterprise businesses access to natural language processing (NLP) powered by large language models (LLMs). “As soon as they see their competitors innovating, they will have to keep up or fall behind,” he said.

Not surprisingly, an army of service providers is lining up to help enterprise companies develop and take advantage of generative AI capabilities.

Consulting firms are pouring money into generative AI efforts

For example, the world’s biggest consulting firms are pouring money into the effort: Bain & Company was first out of the gate to announce a partnership with OpenAI in mid-February. Bain’s highly-publicized work with Coca-Cola on the brand’s Create Real Magic campaign paid off big as Coke highlighted it in its first quarter earnings release this week. 

And Deloitte recently announced a new practice dedicated to helping clients “harness the power of generative AI and foundation models to exponentially enhance productivity and accelerate the pace of business innovation.” 

Finally, PwC announced plans this morning to invest $1 billion in generative AI technology in its U.S. operations over the next three years, including working with Microsoft and OpenAI. Those efforts will include advising clients on how best to use generative AI, while helping them build those tools. 

In an interview with VentureBeat, Mohamed Kande, vice chair of U.S. consulting solutions, co-leader and global advisory leader at PwC, said organizations are excited to use generative AI in their businesses for productivity and other improvements. But they are concerned about the collateral damage they have to manage, such as risks to data privacy. 

That’s why a big chunk of that investment will go to building up PwC’s own generative AI capabilities and expertise, he explained. “We really believe that whatever we recommend for clients to do when it comes to adoption and scaling of technology, we do it ourselves first,” he said. “Then we can say, here are all the lessons learned from it, here’s how we are protecting data.”

Enterprise companies have to be careful with generative AI 

While LLM tech has been in development for the last five to seven years, it is still new from an enterprise deployment context, Rohit Gupta, founder and CEO of Auditoria.AI, an AI-driven SaaS automation provider for corporate finance, told VentureBeat by email. 

“Enterprises are not yet equipped with a consistent evaluation methodology for LLMs, and the ability to quantify ROI on such investment is still work in progress,” he explained. “Also, to leverage the power of LLMs, you need to have it run on your enterprise data, and companies are not yet comfortable opening that up broadly — there will be additional data controls needed.” 

That means that for large enterprises, adopting generative AI isn’t just about logging onto the internet and prompting ChatGPT like consumers do. 

Kande said companies have to be “intentional,” understanding not only how they manage the data they want to use, but the risk within the organization. “We tell them it’s not going to happen in a day,” he said. 

On the other hand, not all use cases are risky, he pointed out. “Some of it is actually good for productivity improvement, without creating any collateral damage,” he said. 

It is the newness, and the distributed nature of the power of generative AI, that is causing many to pause, said Drayton Wade, head of product operations and strategy at AI automation platform Kognitos. But it is being and can be used safely in organizations today, particularly when it comes to automation. 

“When combined with a deterministic, logical system it can be used immediately to drive huge productivity gains safely,” he said, adding that executives should look for generative AI-based platforms with a human review step, full auditability — in natural language — and privacy systems. 

Even ChatGPT is being prepared for enterprise use

As generative AI providers look to take advantage of the enterprise market, even ChatGPT looks like it will be in the mix.

An OpenAI blog post yesterday said that the company is “working on a new ChatGPT Business subscription for professionals who need more control over their data as well as enterprises seeking to manage their end users. ChatGPT Business will follow our API’s data usage policies, which means that end users’ data won’t be used to train our models by default. We plan to make ChatGPT Business available in the coming months.”

But OpenAI competitor Cohere, which specializes in custom, bespoke LLMs, doesn’t believe that offering will meet enterprise needs.

“I’m sure it’ll be a great product,” said Cohere’s Kon, but he cautioned that for mission-critical enterprise use cases, enterprises won’t want to use “generic, standard tools that everyone uses, you want to have a competitive advantage,” he said. “So, by definition, you need to develop these kinds of things based on your own LLM capability, in your data environment, with your proprietary data.” 

Getting over the fear and moving towards AI’s future

While many top enterprise companies are already fully on board the generative AI train — Walmart, for example, recently confirmed to VentureBeat that it is building capabilities on top of OpenAI’s GPT-4 — others have to get over the fear that accompanies the excitement. 

“The reaction you get in Italy, about them banning ChatGPT, it’s out of fear about how to protect the data,” said PwC’s Kande. “We personally believe that the technology exists — don’t fear it, but manage the risk.”  

And that starts, he added, with PwC developing its own generative AI capabilities to pass on lessons learned to clients about delivering on outcomes. “It changes the nature of the discussion to our clients because they’re not just intellectual [discussions],” he said. “They are very practical discussions that we’re having.” 

Originally appeared on: TheSpuzz