Ensuring productive partnerships between organizational end-users and increasingly intelligent software tools is critical to generative AI strategy success. Expect a rocky relationship in need of coaching and coaxing.
In the rush to establish technical strategies for making good on the promise of generative AI, many CIOs find themselves running headlong into what may be their most challenging task yet: preparing their organization’s end-users — from knowledge workers and assembly line laborers to doctors, accountants, and lawyers — to co-exist with generative AI.
Although many analysts, thought leaders, vendors, and chief executives view and position large language models (LLMs) and tools such as Microsoft Copilot as assisting rather than replacing workers, the flood of generative AI products that have hit the market and speedy implementation of LLMs in production to perform many human tasks has challenged that argument, outlining a complex relationship between artificially intelligent machines and the humans who must work with them.
Given the disruptive potential of generative AI, the stakes are high, as Reuven Cohen, strategic AI advisor to Baxter International, a Fortune 500 company, points out.
“The struggle is augmenting your workforce or replace it altogether,” he says. “Step one is likely a question of empowering the most capable people in your organization with highly tailored AI; the next step is removing the less capable altogether.”
But what will be defined as ‘less capable’ will likely be impacted by the technology’s evolution, as well as the evolution of human-machine partnerships wherever the technology is implemented. After all, one increasingly common adage is that generative AI will not replace humans but “humans who are using generative AI will replace humans who are not using generative AI,” as Teradyne CIO Shannon Gath said on a September episode of CIO Leadership Live.
For now, most CIOs are deploying generative AI to enhance productivity and efficiency. Gartner pegs this number at 77% of CIOs. Jamie Holcombe, CIO of the United States Patent and Trademark Office, is one.
“I consider AI an augmented intelligent tool. I don’t feel you collaborate with a tool — you use it,” Holcombe says. “Our examiners are welcoming the help from AI tools to take away the clerical and administrative functions so they can focus more on thoughtful analytics that cannot be merely programmed.”
Consequently, one of the CIO’s top priorities for 2024 is discovering and uncovering the added value human workers can concretely achieve using LLMs — and much of this remains unknown.
Mike Mason, chief AI officer at Thoughtworks, believes CIOs must remain critical when considering new generative AI tools for their workforces in light of this issue.
“Even as AI becomes more advanced and integrated into software and everyday tasks, an influx of AI tools is causing confusion among employees,” he says. “CIOs must remember that it is their workforce who will be using this AI technology, and consider the impact of AI on their workforce, ensuring proper management, training, and integration to make the most of their investment.”
The makings of a close partnership
Even as industry luminaries call for caution on AI, most corporate giants, including Goldman Sachs, Fidelity Investments, Procter & Gamble, American Express, Gilead Sciences, and many others, have gone public about developing and deploying LLMs internally to boost productivity and innovation.
At Fidelity, early returns are proving fruitful for cost savings and increased efficiencies, said Vipin Mayar, the finserv’s head of AI innovation, at the Chief AI Officer Summit in Boston in December.
While he acknowledged LLMs are not on par with human intelligence, Mayar sees the pace of innovation in generative AI as unparalleled. “It’s only been 13 months and it made time nonlinear,” he quipped.
Still, to ensure workers gain the most out of the tools, Mayar suggested multimodal LLMs combining structured datasets and unstructured data should be designed smaller and for specific tasks.
Yvonne Li, vice president of AI, data engineering, and decision science at Advanced Auto Parts, agrees that the technology — and how humans will make use of it — is still in its early stages.
“Gen AI is not a magic bullet,” she said at the summit. “Gen AI can pull data together and gives data scientists a different lens, but it can’t ideate for us. People are using gen AI for efficiency and as a tool for diagnostic problems.”
Thomson Reuters is one organization targeting gen AI for efficiency. The company recently released a generative AI platform that enables legal editors using its Westlaw service to produce document summarization of legal research in minutes that used to take days or weeks to complete, says Shawn Malhotra, head of engineering at Thomson Reuters.
Thomson Reuters’ legal drafting with Microsoft Copilot, another aspect of the platform, unlocks higher functionality for legal editors as well. But observers say innovations such as these will require CIOs to develop upskilling and governance strategies to ensure employees can benefit from new generative AI implementations, wherever they reside. This is fast becoming crucial, as the push for productivity gains is putting pressure on workers across the enterprise to learn to collaborate with LLMs, many of which remain in pilot testing.
“LLMs in many ways can and will exceed human capabilities, but I’m a firm believer that AI will continue to augment humans,” says John T. Marcante, US CIO in residence at Deloitte, and former global CIO at Vanguard. “I think AI will be man’s very close companion now and in the future.”
To ensure an amicable relationship, Marcante stresses the importance of considering stakeholder workflows when implementing generative AI.
“It’s important to remember that using AI to accelerate an outdated or onerous process could be the wrong answer. More benefit may come from a process or technology improvement instead of broad application of AI to ‘fix’ problems,” he says.
Changing how work gets done
Evolutions in the technology, as well as its use, are sure to transform how workers make the most of the tools over time.
At CES this week, Accenture released a public statement that generative AI tools are more “human by design,” pointing to refined conversational user interfaces, robots that respond to English commands, and software that augments how humans work naturally, such as Adobe Photoshop’s Generative Fill and Expand features.
Late last year, Gartner kicked off its annual IT Symposium/Xpo detailing how generative AI is revolutionizing the human-machine relationship.
“It’s more than just a technology or a business trend. It really is shift in how we interact with machines,” said Mary Mesaglio, a Gartner analyst. “We are moving from what machines can do for us to what machines can be for us. Machines are evolving from being our tools to becoming our teammates. “
Machines are not only evolving into work partners but also into customers, Mesaglio said. For instance, connected to a service that monitors usage levels, HP printers are capable of purchasing ink when needed. Tesla automobiles are also capable of ordering parts when a self-diagnosis surfaces a malfunction.
USPTO’s Holcombe also believes that evolutions in interfaces will help workers be more effective with the tools, with the next iteration of human-to-machine interface being audio with natural language rather than keyboards and the mouse. But he still does not see LLMs replacing human cognition any time soon.
“Human thinking and analysis have not been overtaken by machines because the algorithms themselves are at best iterations and trial and error for guessing,” he says. “I’ve never seen a machine make an intuitive leap without it being programmed by a human.”
Usama Fayyad, executive director of Northeastern University’s Institute for Experiential AI, sees conversational AI becoming increasingly important in the enterprise, providing more substantial answers to questions over time. Content generation, document summarization, as well as enhanced analysis and insight extraction tools and decision-making algorithms that require human augmentation will also be important use cases for enterprises across industries, he says.
But for these tools to reach their full potential, how — and how often — they are put to use by humans is important. Such is the nature of the technology.
Joe Atkinson, chief products and technology officer at PwC US, sees generative AI applications helping to create a more tech-savvy workforce. But it remains unclear how workers will add value to the tools themselves, which, by design, learn as they go. No doubt human creativity will be necessary to elevate the quality of application, he says.
To that end, Gartner advises CIOs to establish “lighthouse” principles that define how workers and machines will interact in the year ahead — a priority that the firm puts on par with making data AI-ready and implementing AI-ready security.
After all, generative AI is not a set-it-and-forget-it tool — at least not yet. It requires human oversight and experience to assure accuracy, quality outcomes, and safety.
As part of this push, CIOs are gearing up with education and training sessions, implementing generative AI tools into the workplace gradually, and reassuring workers that AI tools are designed to augment their work and not replace them.
Sreenivasan Narayanan, executive vice president of Nous Infosystems, an enterprise technology consultancy in Dallas, has attended an AI program at the Wharton School of Business and has trained 42% of Nous’ workforce on Level 1 AI skills.
“We were dabbling with GitHub, PowerApps, Teams, M365, and Security Copilots so far in our digital labs a while back,” he says. “In the last few months, we have deployed this to production-grade client environments to provide solutions around code generation, case document summarization, voice answering, language translation,” he adds. “The workforce will embark on Level 2 [training] while more inducted in this organizational transformation.”
The human factor
But not all are taking their employers’ word for it.
Microsoft and the AFL-CIO recently announced the creation of a partnership, described as a first of its kind, designed to keep the dialogue open about AI development and how it may impact workers’ needs and roles, incorporate worker feedback, and shape public policy that supports the technology skills and needs of frontline workers, according to Microsoft.
And at its IT Symposium, Gartner led off with what it says was an unusual, but necessary call to arms: that machines are taking on different roles, and in some cases, human roles, and this cannot be ignored.
Moreover, the rapid pace of innovation of ChatGPT and development of capabilities such as DocLLM — which would be far more accurate in extracting data that is unstructured, such as images and video — has some wondering whether human-like artificial capable intelligence (ACI) and artificial super intelligence (ASI) will arrive sooner than expected and alter the value equation in the machine’s favor.
In the meantime, the daily evolution of generative AI platforms is exciting to developers and eagerly awaited by enterprise executives. For CIOs and CTOs, it’s a balancing act of cost vs ROI. Generative AI solutions are expensive to build and deploy and that will temper enterprise adoption, observers say.
“As CTOs, we need to work on quickly evaluating new tech, and whether it makes sense for our companies and what we need for our users,” says Jeremy King, SVP and CTO of engineering at Pinterest. “It’s a lot to evaluate — from whether to ‘buy or build’ to ensuring it works with existing foundations.”
Chief among those foundations — at least for now — is the company’s workforce. CIOs should strategize accordingly.
>>> Read full article>>>
Copyright for syndicated content belongs to the linked Source : CIO – https://www.cio.com/article/1290353/cios-confront-generative-ais-workplace-x-factor.html
