Analyst firm Gartner says artificial intelligence and its major developers are not yet ready for enterprises to use the technology. This was announced at the annual Gartner IT Symposium by leading analysts and company vice presidents Daryl Plummer and Kristin Moyer. They said the AI industry and its key players have not yet reached the maturity required by the enterprise sector, writing register.
When it comes to working with leading artificial intelligence labs, Plummer said it’s too early to trust them. According to him, they “Does not meet company standards, does not understand company terms and conditions, does not understand liability issues, and does not understand concepts such as addressing the stability and continuity of business processes.” Plummer noted that it’s common practice for AI companies to change models frequently, keeping in mind how updates might impact applications that rely on their system’s performance and break the software.
Image credit: Look Studio/Unsplash
“The pace of innovation is out of control and sadly it cannot be ignored,” – Plummer noted. Making the problem worse, he said, is that AI developers are not prepared to support outdated versions of AI models, even though their life cycles are only about six months. “If you decide to stick with the first version of the model, you’re not going to get a lot of attention. That alone shows they’re not ready to work with the enterprise sector.” – he emphasized.
Christine Moyer cited a study by Gartner that showed that 86% of information chiefs (CIOs) believe that the risks associated with artificial intelligence are growing faster than the benefits it brings. In particular, the early success of AI adoption is creating needs that IT departments cannot meet without compromising system security and stability. Moyer attributes this demand in part to “irresponsible consumption,” where workers use artificial intelligence unnecessarily or inappropriately. Research shows that 40% of employees have experienced low-quality content generated by artificial intelligence (AI slop). It takes about two hours to parse and understand, which can cost $9 million per year for an organization with about 1,000 employees.
It is extremely difficult to put an end to this kind of “irresponsible consumption” because “It’s even hard to track exactly how many AI agents you’re running. AI is coming into your company through the products you’re already using.”Moyer said. She noted that it took a decade for enterprises to prevent developers from choosing the most powerful and expensive cloud instances among those available.
Photo credit: Nahrizul Kadri/Unsplash
Plummer believes that time-tested automation techniques can be used to control the use of artificial intelligence. According to him, it’s unwise to create a proxy to interact with the repository, since a good old function call will do the job just fine. Part of the blame for irresponsible use of artificial intelligence lies with the vendors “They are working hard to monetize artificial intelligence at all costs,” By selling products under its auspices or simply forcing customers to buy more tokens. “Vendors want you to use a proxy everywhere,” – said Plummer.
When it comes to the effectiveness of AIOps, Plummer said vendors are moving in the right direction, but they are driven by nefarious motives. “Their goal is to sell you their product rather than implement a truly appropriate solution; after all, their product probably isn’t,” analysts think. According to him, this behavior is typical of new technology sales cycles; it’s happening as vendors create management tools that they say can safely use artificial intelligence. Plummer described the market for such solutions as “one of the most fragmented we have ever seen.”
Analysts have proposed a number of measures to “contain” artificial intelligence, one of which is the creation of an “artificial intelligence central bank.” This structure will be responsible for monitoring the use of AI across the organization and ensuring its stable operation, taking into account the systemic impact of AI. Its key task is to ensure accountability. This can be accomplished using a traditional ERP or CRM system that records the history of jobs (who performed or authorized those actions). Artificial intelligence doesn’t always leave such a mark, Moyer added.
Image credit: Francis Ledger/Unsplash
Furthermore, in order to eliminate systems that have become out of control, “custodial agents” will be needed – artificial intelligence whose task is simply to monitor other agents and ensure that their behavior remains within specified limits. He should also be given the power to “eliminate” runaway agents. Additionally, experts recommend establishing rapid response teams to deal with the consequences of glitches or problems caused by AI. “If you have to take responsibility for something that happened and you say ‘it was the AI,’ then you’re in big trouble.” Plummer said. — The chief information officer will be responsible for the failure of the artificial intelligence system. “
source:
