When most of us think of artificial intelligence in the workplace, we imagine automated assembly lines of robots managed by an algorithm. LinkedIn's Reid Hoffman has a different idea.

In an essay for MIT Sloan Management Review, Hoffman describes human applications for the technology. Among other things, it would help to use data science to improve the way we onboard new team members, organize workflow, and communicate about performance. Addressing the question of how technology will change management practices over the next five years, Hoffman explains how the use of a "knowledge graph" will become standard management practice.

Imagine a spider chart mapping the complex web of interactions and data transfers within your company. A.I. can process large amounts of files, emails, and data providing unprecedented access to information about how we delegate work, tackle assignments, and deliver results.

This kind of oversight can sound intimidating, and many skeptics worry about invasion of privacy and worker displacement with the rise of A.I. But Hoffman, like tech-industry colleagues including Sam Altman, Elon Musk, Peter Thiel, and Jessica Livingston, is a proponent of the OpenAI project, which seeks to develop technology that will help, not harm, humanity. Here are three ways he thinks A.I. could help your office run more smoothly.

1. Faster onboarding

New hires spend weeks getting up to speed--meeting with new colleagues, reading through files, and navigating internal processes. Hoffman says a knowledge graph, full of information gathered before the new employee was hired, can answer questions like "Who do I need to work with on the new office move? What were the meetings where it was discussed? When is our next status meeting?" Simple questions, perhaps, but with dozens of such hurdles and no knowledge of company processes, a new employee would save time having a digital repository for the minutiae of orientation.

2. More efficient workflow

Forget emailed meeting recaps. A.I., as Hoffman sees it, will process decisions and assignments made in the conference room and integrate them into the knowledge graph. He points out that the technology for classification and pattern matching already exists, as demonstrated by Google's search box autocomplete and Amazon's product recommendations. In addition to linking decisions with assignments and steps needed to complete a project, Hoffman thinks A.I. has the potential to measure less tangible workplace influences. He envisions a dashboard that combs internal communication to assess workplace sentiment. Knowing what issues are being discussed most frequently, what concerns are most often analyzed, and how financial and emotional capital are used could provide a valuable layer of insight for managers. Hoffman credits Joi Ito, director of the MIT Media Lab, for the term "extended intelligence," or "treating intelligence as a network phenomenon and using A.I. to enhance, rather than replace, human intelligence."

3. Objective performance reviews

Hoffman cites a staggering figure from a Deloitte study that found only 8 percent of organizations said annual performance reviews were worth the effort they require. While some jobs are easier to measure--widgets sold, new customers acquired--what counts as success in most positions is more subjective. Differing opinions, workplace politics, and unconscious bias take their toll on performance reviews. A knowledge web could capture every tiny detail of who proposed a key idea in a long-ago meeting and who managed the tasks to make it happen. "If performance management were a movie, it will become less Gladiator, and more Moneyball," Hoffman writes. To be clear, he doesn't see an end to the need for people skills at work and human judgment in reviews, but he argues A.I. can help managers objectively identify patterns in workers' strengths and weaknesses. That can then help managers make assignment decisions.