Using Artificial Intelligence to Humanize Management and Set Information FreeUsing Artificial Intelligence to Humanize Management and Set Information FreeUsing Artificial Intelligence to Humanize Management and Set Information Free

Reid Hoffman

This essay originally appeared on MIT Sloan Management Review as part of their Frontiers Essay Series. Each essay is a response to this question: “Within the next five years, how will technology change the practice of management in a way we have not yet witnessed?”

Artificial Intelligence is about to transform management from an art into a combination of art and science. Not because we’ll be taking commands from science fiction’s robot overlords, but because specialized AI will allow us to apply data science to our human interactions at work in a way that earlier theorists like Peter Drucker could only imagine.

We’ve already seen the power of specialized AI in the form of IBM’s Watson, which trounced the best human players at Jeopardy!, and Google DeepMind’s AlphaGo, which recently defeated one of the world’s top Go players, Lee Sedol, four games to one. While these specialized AIs can’t pass a Turing test, they can process and manipulate enormous quantities of data at a rate our biological brains can’t match. Herein lies the parallel to management: Within the next five years, I expect that forward-thinking organizations will be using specialized AIs to build a complex and comprehensive corporate “knowledge graph.”

Just as a social graph represents the interconnection of relationships in an online social network, the knowledge graph will represent the interconnection of all the data and communications within your company. Specialized AIs will be ubiquitous throughout the organization, indexing every document, folder, and file. But they won’t stop there. They’ll also be sitting in the middle of the communication stream, collecting all of our work product, from emails to file shares to chat messages. They’ll be able to connect when you save a proposal, share it with a colleague, and discuss it through corporate messaging. This may sound a bit Big Brother-ish, but the effect will be to give knowledge workers new and powerful tools for collecting, understanding, and acting on information.

Specialized AI will even help us improve that scourge of productivity, the meeting. Meetings will be recorded, transcribed, and archived in a knowledge repository. Any time someone in a meeting volunteers to tackle an action item, AIs will record and track those commitments, and automatically connect the ultimate completion of that item back to the original meeting from whence it sprang. Sound far-fetched? The AI techniques for classification, pattern-matching, and suggesting potentially related information are already part of our everyday lives. You encounter them every time you start typing a query into Google’s search box, and the autocomplete offers a set of choices, or every time you look at a product on Amazon, and the site recommends other products you might like.

The rise of the knowledge graph will affect the practice of management in three key ways:

1. Meaningful organizational dashboards

Right now, organizational dashboards — the sets of information executives monitor and use to guide decision making — are limited to the structured data that is easy to extract or export from existing systems — things like revenues, app downloads, and payroll information. These backwards-looking metrics do have value: They help managers understand what has happened in their operations and identify hotspots for troubleshooting. But AI-generated knowledge graphs will dramatically expand the scope of these dashboards. For example, managers will be able to access sentiment analysis of internal communications in order to identify what issues are being most discussed, what risks are being considered, and where people are planning to deploy key resources (whether capital or attention). AI-powered dashboards will provide forward-looking, predictive intelligence that will deliver a whole new level of insight to managerial decision making. Computers won’t be making decisions for us, but they can sift through vast amounts of data to highlight the most interesting things, at which point managers can drill down using human intelligence to reach conclusions and take actions. This is an example of what Joi Ito, director of the MIT Media Lab, refers to as “extended intelligence” treating intelligence as a network phenomenon and using AI to enhance, rather than replace, human intelligence.

2. Data-driven performance management

Current performance management processes are terribly flawed. A Deloitte study found that just 8% of organizations believe that their annual review process is worth the effort! One of the big reasons for this dissatisfaction is the lack of data to drive objective performance management. In order to manage performance, you have to be able to measure it, and in most organizations, this simply isn’t possible for the majority of their people. Senior leaders might be evaluated based on the company’s overall performance, and certain functions like sales have objective, quantitative performance metrics, but almost everyone else is relying on subjective criteria and analysis. In the absence of data, internal politics and unconscious bias can play a major role, resulting in performance management that is biased and inaccurate. The knowledge graph will allow managers to identify the real contributors who are driving business results. You’ll be able to tell who made the key decision to enter a new market, and which people actually took care of the key action items to make it happen. If performance management were a movie, it will become less “Gladiator,” and more “Moneyball.” Yet even as the knowledge graph reduces the role of guesswork and intuition, the human manager will still be in the loop, exercising informed judgment based on much better data. The result will be much more efficient allocation of human capital, as people are better matched with projects that suit their strengths, and the best people are deployed against the highest-leverage projects.

3. Increased talent mobility

As we get better at allocating human capital, organizations will need to do a better job of supporting increased talent mobility, both inside and outside the organization. In the networked age, talent will tend to flow to its highest-leverage use. Each such “tour of duty” will benefit both the company and the worker. But people are not plug-and-play devices; they need time to become productive in a new role (largely because it takes time to build the needed connections into a new network). The knowledge graph will make onboarding and orientation far more rapid and effective. On the very first day on the job, the worker will be able to tap the knowledge graph and understand not just her job description, but also the key network nodes she’ll need to work with. Rather than simply being handed a stack of files and being introduced to her colleagues, an onboarding AI will be able to 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?” She’ll also be able to ask how things were done in the past (i.e., “Show me a tag cloud of the topics my predecessor was spending his time on. How has that allocation evolved over the past 12 months?”). AIs might even ask outgoing employees to review and annotate the key documents that should be passed on to their successors. The tacit knowledge that typically takes weeks or months to amass in today’s workplace will have been captured and processed in advance so that within the first hours of accepting a new role, an employee will be able to apply that knowledge.

For all its potential benefits, some very smart people are worried about the potential dangers of AI, whether they lie in creating economic displacement or actual conflict (such as, if AI were to be applied to weapons systems). This is precisely why I, along with friends like Sam Altman, Elon Musk, Peter Thiel, and Jessica Livingston, are backing the OpenAI project, to maximize the chances of developing “friendly” AI that will help, rather than harm, humanity. AI is already here to stay; leveraging specialized AI to extend human intelligence in areas like management is one way we can continue to progress towards a world in which artificial intelligence enhances the future of humanity.

This post was originally published here on June 28, 2016