By W.B. King
While the first pieces of Halloween candy have yet to be gobbled up, Info-Tech Research Group has already released its forecasting report, Tech Trends 2025: Big Risks. Bigger Possibilities. Among findings are how organizations should be viewing a handful of trends, including exponential artificial intelligence (AI) initiatives and the rise of digital humans.
“As firms push their investment into AI and more specifically generative AI, the focus shifts to simulating the future,” said report author, Brian Jackson, director at Info-Research Group. “CIOs must push to develop forecasts and advance the knowledge of their organizations in a way that carefully manages the integrity of the knowledge creation and curation process. Like the Manhattan Project, CIOs need to create a trusted environment that ensures ‘epistemic security.’”
Since 1997, the London, Ontario, Canada-based firm has provided IT research designed to help CIOs and IT leaders make strategic, timely and well-informed decisions. Between March and July 2024, 970 IT decision-makers from disparate industries were surveyed.
Exponential AI Initiatives
Among those polled, 80% of “transformers” said they are already invested in exponential AI or will be by the end of 2025, compared to 72% of “average” respondents.
“For the most part, the investment is already made – only slightly more than one-quarter of all organizations say they aren't invested yet, but plan to invest by the end of 2025,” the report noted. Organizations bullish on AI see it fitting into the next wave of digital transformation. It can augment many different business processes, and it also promises to upend some business models entirely, demanding new ways to interact with customers and likely raising expectations even higher.”
According to the report, integrating AI solves business process operational challenges, including architecting a high-quality and specialized data pipeline that can be used to fine-tune and pretrain foundation models and navigate myriad tech platforms and interoperability issues.
“Those persistent enough to solve these problems will reap the rewards. As AI crests the wave of technology adoption and pushes through the emerging phase and into the transformative phase, best practices to unlock value will become more evident,” Jackson noted. “Those riding the leading edge of that wave demonstrate some of the best ways to see that return on investment made as early as possible.”
As an example of generative AI benefits, Jackson pointed to SAS’ data analytics platform SAS Viya, which recently assisted a banking client in applying natural language processing (NLP) to customer complaints received across various channels. The Cary, N.C.-based fintech SAS has more than 40 years of experience in data analytics and AI. Marinela Profi, strategic AI advisor at SAS, offered that while AI is an important tool, the human touch can’t be lost.
“A lesson learned so far is that LLMs alone don’t solve business problems. You need humans and a platform that enables the integration of LLMs in the business’ existing systems in safe ways,” Profi noted in the report. “In the customer complaints use case, the LLM only generates the reply of the complaint. The end-to-end process is enabled by a governance platform.”
Turing Test Challenges
In 1950, computer scientist Alan Turing developed a protocol to tests a machine's ability to exhibit intelligent behavior equivalent, which today is known as The Turing Test. A recent study at UC San Diego confirmed just how difficult it has become to distinguish humans from AI.
“Participants were asked to have a five-minute text-based conversation with another party and then guess if it was a human or a machine. Participants correctly identified humans 67% of the time, but OpenAI's ChatGPT models also did quite well,” the report found. “ChatGPT 3.5 was identified as human half the time, and GPT-4 was identified as human 54% of the time. In other words, being able to identify a human from a machine was no better than chance.”
The good news is that generative AI chatbots facilitate good experience when interacting with users like they would interact with their co-workers, conversationally. “They cut through arcane icon and menu interfaces that sometimes represent hundreds of different features, providing a user exactly what they ask for,” the report noted. “They are particularly good and finding relevant knowledge and summarizing it and can sometimes even generate exactly what the user needs to complete a task.
Where generative AI goes wrong is that it is prone to errors and can be biased by training data—these factors could lead to risks around data confidentiality and other performance problems, the report stated. Even still, the global AI voice generator market is estimated to grow at a rate of 15.4% per year from 2022 to 2032, with a projected value of almost US$1.9 billion in 2025 and US$4.9 billion by 2032, the report said.
When it comes to using generative AI chatbots (now or in the future), 78.84% of respondents said they use Microsoft Copilot, while 43.48% noted ChatGPT Enterprise. Other platforms used include Open Source, Google Gemini/Google Duet and Salesforce Einstein Copilot.
“Most organizations are using at least one chatbot or are planning to deploy one or selecting one. Many of the options are easy to adopt, offered as upgrades to existing software packages or as simple subscription services for a web service,” the report noted. “For example, Microsoft Copilot is a chatbot integrated with the software firm's Office 365 productivity suite. It dominates the field, with nearly eight in 10 chatbot users saying they use it or plan to use it.”