CIOs press ahead for gen AI edge — despite misgivings

OpenAI’s November 2022 announcement of ChatGPT and its subsequent $10 billion in funding from Microsoft were the “shots heard ’round the world” when it comes to the promise of generative AI.

If anything, 2023 has proved to be a year of reckoning for businesses, and IT leaders in particular, as they attempt to come to grips with the disruptive potential of this technology — just as debates over the best path forward for AI have accelerated and regulatory uncertainty has cast a longer shadow over its outlook in the wake of these events.

But if there are any stop signs ahead regarding risks and regulations around generative AI, most enterprise CIOs are blowing past them, with plans to deploy an abundance of gen AI applications within the next two years if not already. Of roughly 2,500 CIOs surveyed recently by Gartner, 9% say they have already deployed gen AI applications, and a staggering 55% say they will deploy large language models (LLMs) in production by the end of 2025.

“We’re seeing lots and lots of pilots,” says Gartner AI analyst Arun Chandrasekaran, who notes content creation, document summarization, sentiment analysis, and enterprise search chief among the initial use cases.

A recent survey of nearly 1,000 IT decision-makers conducted by Foundry underscores this. Of the 92% of IT executives expressing interest in gen AI, 57% are busy identifying use cases, 45% are launching pilot programs, 41% are training employees on gen AI, and 40% are establishing policies and guidelines. Moreover, four out of every five IT executives plan to leverage gen AI to augment and enhance developers’ productivity, with more than 40% of those IT leaders expecting to do so within the next year.

Innovative CIOs from CarMax, Goldman Sachs, Coca-Cola, L’Oréal, Georgia-Pacific, Nasdaq, and Vodaphone are among those whose large enterprises have announced publicly that they are in pilot mode — and in a few cases, deploying production workloads — with LLMs. OpenAI’s late August announcement of the release of ChatGPT Enterprise based on GPT-4 included note of its use by Estée Lauder Cos., PwC, Zapier, Block, Canva, and Carlyle.  

Overall, as we near the first anniversary of ChatGPT’s release, the business world appears to be one in transition, working to lay the foundation for new gen AI–infused capabilities and platforms, with CIOs capably in the spotlight — and early misgivings, while not entirely forgotten, increasingly upstaged.

Gen AI boom in the making

Many early and established forays into generative AI are being developed on the AI platforms of cloud leaders Microsoft, Google, and Amazon, reportedly with numerous guardrails and governance measures in place to contain unrestricted exploration. 

CarMax CITO Shamim Mohammed confirms his company was using OpenAI’s GPT-3.x natural language model on pilots well before Microsoft’s highly publicized OpenAI investment and is moving full steam ahead on streamlining content creation for car research pages and encouraging its programmers to scale use of GPT-4 to further enhance the customer experience.

Mercedes-Benz is also leveraging gen AI, in its case to elevate in-car voice control to enhance drivers’ experience through its use of Microsoft Azure OpenAI Service, which enables IT organizations to parlay the power of OpenAI’s GPT-4 and GPT-3.5 in concert with Microsoft’s AI-optimized Azure platform.

John Spottiswood, COO of Jerry, a Palo Alto, Calif.-based online insurer, is developing an LLM-based platform to handle customer requests with far more intelligent and enhanced chatbots. “We’re using OpenAI primarily, calling ChatGPT and using a mix of GPT-4 and GPT-3.5. We can also use Claude 2, primarily if the OpenAI API is unresponsive.”

In September, Deloitte AI Institute released a report citing six verticals — financial services; technology, media, and telecommunications; energy, resources, and industrial; consumer; government and public services; and life sciences and healthcare — as ripe for gen AI transformation. 

“As the use of generative AI becomes more widespread, it is causing significant disruption in many industries and sectors,” says Ritu Jyoti, group vice president of AI and automation research at IDC.

Financial security company Apex is employing Genpact’s riskCanvas software, which has been integrated with Amazon’s Bedrock gen AI technology, initially for document summarization but eventually it will be used to help prevent financial crimes, says Justin Morgan, Apex’s head of financial crimes compliance. 

“Apex is among the first to implement this into production,” Morgan says. “There aren’t many examples of gen AI in production out there, especially at this level.”

Morgan explains that the gen AI is not performing the analysis to determine suspicion of financial crimes. “It’s taking into account the information that the analyst has provided in their initial summary and then coming up with the narrative piece for them,” he says.

Snap, LexisNexis, and Lonely Planet are also developing and training LLM models, each leveraging their own data stored on AWS.

“We are setting up the system using retrieval-augmented generation and prompt engineering alongside Anthropic LLMs on Amazon Bedrock to extract experiences that then fold up into itineraries [formed from] our corpus of print content,” says Chris Whyde, SVP of engineering at Lonely Planet.

Nasdaq, another AWS customer, is using LLMs to “create new kinds of intelligence reports for investors and corporate customers that leverage the company’s proprietary data sets and drive faster, more impactful content creation in Nasdaq’s marketing and communication teams,” says Nasdaq CIO and CTO Brad Peterson, who also noted in a recent interview with CIO.com that generative AI “represents the next generation of this technology,” and that Nasdaq will use it to expand its portfolio of financial SaaS products for identifying financial crime risks and enabling corporate boards to produce and consume presentations and required disclosures more efficiently.

Tom Richer, head of Google Business Group at technology services company Wipro, is also seeing rapid traction among CIOs eager to explore gen AI’s transformative potential. “To this end, many are conducting proof-of-concept projects to assess how generative AI can be leveraged to streamline IT operations, enhance data management, and improve decision-making,” he says.

The WiPro exec cited an example of a recent POC that a life sciences firm conducted using Google’s Gen App Builder to reduce contract management costs and improve agreement adherence to company policies.

“The prototype generated supplier agreements by leveraging structured prompts fed by the contract manager in adherence to policy,” Richer says. “The POC resulted in significant time savings for contract generation, clause searching and summarization, and overall accuracy in the contract management function.”

Meanwhile, established AI vendors are also moving quickly to offer advanced gen AI platforms to keep their customers close. Databricks, for instance, purchased MosaicML to enable its customers to employ LLMs in their AI applications. 

C3 AI, another established AI vendor, last month released an enhanced Generative AI Suite with 28 domain-specific gen AI offerings. According to C3, sugar producer Pantaleon is using C3 Gen AI to supplement sales forecasting, while Georgia-Pacific is using it for manufacturing process knowledge. Additionally, Flint Hill Resources is deploying the LLM-based platform for commodity trading optimization, while the US Missile Defense Agency is employing it to improve safety during steel manufacturing, according to C3.

Balancing risks, rewards

The rate of pilot testing and POCs — this early in the game — is quite high, particularly for a rapidly advancing technology deemed by Elon Musk and others as potentially “civilization” destroying. Musk and other technology luminaries earlier this year signed an open letter calling for a six-month pause in the frenzied race in AI development.

In his book The Coming Wave: Technology, Power, and the Twenty-First Century’s Greatest Dilemma, Mustafa Suleyman, co-founder of DeepMind (owned by Google) and now CEO of Inflection AI, warns about the combination of more advanced generative AI with synthetic biology. In a recent podcast, he further explained that generative AI could evolve into an “artificial capable intelligence” that will have full agency to make decisions independently of humans and at a scale far beyond what humans can comprehend.

To guard against this, Suleyman and others advocate for human involvement in strictly enforcing responsible use through frameworks and laws, much the way nuclear arms and biological weapons are managed. 

Yet, the intense focus on gen AI has only accelerated experimentation for CIOs and vendors, including Musk, whose xAI will reportedly enter the AI arms race.

“The companies I talk to are not worrying about whether they’ll create an evil genius AI that destroys the world,” says George Westerman, a senior lecturer at MIT Sloan School of Management. “They’re just trying to solve the business challenges they face right now.”

And for most CIOs that means enhancing productivity. Nearly 60% of respondents to Foundry’s AI Priorities Study 2023 agreed that gen AI will play a large role in employee productivity, with 55% saying that gen AI–infused products produce better business outcomes.

Power supply giant Generac is one company that’s all in on gen AI, says CIO Tim Dickson.

“We are now fully embracing generative AI, with three innovative pilots that are live,” he says. “First, we launched a private instance of GPT-3.5 for internal enterprise exploration. Next, we launched a customer service chatbot to answer customer call questions for our customer service reps. Lastly, we tapped into our data lake to enrich and tailor specific customer emails to drive the conviction of our products and ultimately increased sales. These three programs are already delivering value for the business.”

And doing so requires taking risks, he says, something he believe IT leaders must embrace to succeed today. “We are indoctrinating a culture of gen AI within the company,” he adds.

Still, the widening availability of gen AI to the public at large keeps many CIOs awake at night. Few enterprises have slammed the brakes, but no doubt it has led to a high emphasis on corporate guardrails, frameworks, and shared responsibility in the C-suite.

Security and privacy remain key concerns for IT leaders, according to Foundry’s AI survey, as are authenticity, bias, and explainability, with 24% of IT leaders concerned about misuse. While CIOs will be looked to for expertise in addressing these issues, CIOs need to make it clear they are not going to be the fall guys for minor errors, aka AI “hallucinations,” or for more serious consequences to the company and society at large.

“It’s about co-leadership. Not just having a seat at the table but CIOs want the business to be accountable and to share responsibility for the technology outcomes as well as the business outcomes,” says Mandi Bishop, a vice president and analyst at Gartner, who adds that CIOs also want co-ownership and co-delivery of generative AI applications by “fusion teams” that unite business and IT pros to work together in structured units.

The human side of the equation

For most business CIOs, the ley applications for gen AI in the near term are likely to be in customer support, content creation, data augmentation, personalized recommendations, and product design, according to Foundry’s AI Priorities Study 2023.

And for many the inroads will be gen AI enhancements in the platforms they already rely on, as enterprise software vendors increasingly announce new gen AI capabilities for their products.

For example, Teradyne CIO Shannon Gath, who recently participated in a generative AI forum at Northeastern University’s Roux Institute with about 30 Boston area CIOs to discuss AI strategy, says her company is using generative AI incidentally today, through security products with built-in gen AI capabilities, as well as through Bing Enterprise. But she knows that will change, saying in a recent episode of CIO.com’s Leadership Live that generative AI is going to remain a core transformational fabric for the business world and society over the next decades.

“It is very real. Everyone is touching it on a daily basis,” Gath said of her company’s use of gen AI, before adding a twist to a common concern regarding job displacement and gen AI: “Humans who are using generative AI will replace humans who are not using generative AI.”

That emphasis on interaction between human beings and generative AI–based intelligent machines in the near term is what will fuel business value — and prove to be the true disruptive force across industries. And here perhaps resides IT leaders’ impetus for pushing forward despite concerns over where all this will lead.

“Gen AI will be the defining technology of our time and the winners in leveraging this technology will win big,” Lonely Planet’s Whyde says. “The fear of falling behind is a major factor in technology leaders’ minds today.”

Still, as Whyde concedes, “some of the concern around risk of adoption of AI is distressing.” And with this he sums up the tough spot CIOs find themselves in — now and in the years to come.