Discussions about AI don’t get very far before someone mentions keeping a human in the loop. As companies look to a more hands-off approach to AI, they could run into an Amelia Bedelia scenario, where their agentic solution is prone to making larger scale mistakes in the spirit of being a helpful bot.
IT professionals agreed that industry is looking to have less of a human touch when it comes to automating processes, like answering quick questions from employees and customers about IT issues concerning things like accessing an email account.
Pat Casey, chief technology officer and EVP of DevOps at ServiceNow, said that he is seeing more customers ask about taking the human out of the loop for these situations, “and still hav[ing] the constraints in place to prevent the Amelia Bedelia scenario.”
He believes that customers wanting to move away from keeping a human in the loop should put more technical controls into place around the agent, preventing it from stepping out of line or being swayed by bad actors.
Talkin’ bot it. Casey said that the use of AI at a company’s IT help desk can be helpful for deflection rates, but the technology should be given the same level of privilege as a human in the position would, rather than complete access.
“Just like a human being, you can make an honest mistake, you can be badly motivated, you can be suborned by a third party,” Casey said. “We have years and decades of governance in [the] technology world about understanding how to take imperfect human beings and put them in part of a business process and get good, repeatable outcomes.”
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
He continued: “You got to assume agents need to flow through the same imperfect pipeline, rather than through some hypothetical, ‘It never makes mistakes,’ technology pipeline.”
Chief AI officer for industrial AI company IFS, Bob De Caux told IT Brew that he does believe that the human touch with AI bots will be relaxed, but not for some time.
“Even as we remove humans from the loop, it’s not like everything will need to be fully agentic in the way that an LLM will determine the process for decisions to be made,” De Caux said.
AI to rule them all. Vedant Sampath, CTO at Nexthink, told IT Brew that using AI for help desk support and automating the solutions is “giving the power to the employee to resolve the issues themselves,” which could decrease the need for professionals working in these positions.
“Initially, it will drastically reduce the size of the support desk, because the same number of people will be able to answer many more tickets,” Sampath said. “There will be naturally fewer tickets coming to them…and they might be serving the purpose of bumping it up to a [level two] or other person.”
Sampath suggested training employees to use AI in order to move them to a different position, but maintained that a support desk will not be staffed the way that they have been traditionally, or even currently.