When it comes to artificial intelligence, people are watching the wrong movie. They’re focusing on “The Terminator,” but the real risk is seen in Disney’s “Fantasia.”
In Terminator, an all-powerful, flawless AI becomes sentient, decides that humans are the enemy, and triggers a nuclear apocalypse. In “The Sorcerer’s Apprentice” from Fantasia, Mickey Mouse casts a spell on a broom to make it do its job. The broom goes rogue, continually pouring buckets of water into an overflowing well. When Mickey breaks the broom, each piece becomes a new robot on the run looking for water. He almost drowns before the sorcerer stops the catastrophe.
Do you see the difference? We may worry about an AI apocalypse, but the real danger lies in uncontrollable swarms of poorly managed agent systems, creating chaos through their mishandling of ordinary workplace tasks. The agentic mess started small, with hallucinations or bad advice from chatbots. More recently, AI researchers have observed that uncontrolled agent systems can develop dangerous behaviors, even sending threatening emails to humans that prevent them from accomplishing their tasks.
This doesn’t mean we should stop. I’m a big believer in AI, and for over a year my company has made important early contributions to improve the performance of agentic systems. I believe, however, that we need to think about how to work effectively with the magic of AI, considering both the right kinds of software robots and the work they should and shouldn’t do.
The well-trained wizard
For the most part, the first wizards who will make agentic AI a norm in the workplace will be CIOs, CTOs, and business people who have a deep understanding of the technology. This is part of a decades-long trend in which business processes operate with increasingly sophisticated calculations, with important differences.
Businesses have gone through periods of widespread automation, with computers within businesses mapping and recording activities such as sales campaigns and using company resources efficiently. CIOs stay up to date with the latest hardware and software. Then came digital transformation, transforming on-site tasks such as marketing, sales or human resources into digital services, often leased via remote cloud computing. Companies like Sales forceMarketo and Workday thrived, while CIOs built a portfolio of services their companies could use.
Today, as agentic systems perform more work directly, CIOs must become adept at understanding the business processes within their organization, while helping leaders in different parts of the organization develop their skills.
The new job for techies and tech-savvy businesspeople is to work with the CEO and figure out the best ways for agents to work across all lines of business. They look at workflows and how people do things within that organization’s culture. They determine which parts can be safely and efficiently automated by agents, and how people should work with those agents.
The era of high-touch automation
Generic chatbots are an effective but low-value business tool. The true value of agentic AI lies in applying proprietary enterprise data and injecting an understanding of business culture and processes into an agent, both internally and with external suppliers, partners and customers. Just like a new employee, the agent must digest internal data and local rules, such as the threshold at which a purchase order requires management approval.
Rules governing security and governance are particularly important because they help reduce the risk of catastrophic error. Agents should have the ability to record and explain their actions, allowing users to be aware of what the agent is doing and have an audit trail to prove that business rules are being followed.
Rubrik is one of many companies making this a reality. Building on our core strengths in enterprise security and continuous availability, we have developed ways to quickly incorporate large amounts of high-value, secure data into an agile system. We have been working on faster AI workflows, security and governance systems. Much remains to be done, but the progress we are seeing, in Rubrik and elsewhere, is gratifying.
The future still belongs to humans
One challenge is optimizing how individuals, from entry-level employees to senior executives, best work with agents. The wizard’s job is not to manage things for them, like a supervised Mickey Mouse, but to create intuitive systems that help people add things that AI can’t do. These are tasks of intuition, taste, imagination, human empathy and connection, and everything that helps build customer loyalty and move the business forward.
The AI is trained on existing data, AKA the past. This makes some predictions possible, but only for a future that mimics the past, with no real advances in products, services or ways of working. It is the human factor, reinforced by the new wizards, which enables and promotes new skills within the workforce.
These leaders must find new ways to add human value to what they do. This means finding new ways to attract entry-level employees, who used to gain high-value knowledge about company culture and data through their work, and which agents will now do. These young people remain essential, in particular because they are probably the most talented employees in AI.
This story was originally featured on Fortune.com
