Sergey Nivens - Fotolia
Artificial intelligence in HR is likely to transform HR operations in three profound ways.
First is the emergence of the conversational interface, where we can talk to systems, ask questions and interact through chat. This can be supplemented by augmented and virtual reality, which is developing even faster than we thought. Second is machine learning, where software analyzes people-related data and offers smart recommendations and decisions. The third is the growth of predictive models, which are systems that can identify patterns and quickly find areas of risk, fraud and other possible performance problems.
Since most of the non-transactional parts of HR are based on judgment -- virtually every management and people decision is based on imperfect data -- I believe AI in HR has the potential to improve and automate many of the ad hoc decisions we make on a daily basis. The following are some examples:
- Recruiting: There is a wide range of new AI tools used in recruiting. For example, there are AI-based chat systems that can communicate with candidates and quickly screen people. These systems already enable candidates to select the right job or shift, and they can dramatically reduce the time recruiters spend on candidate screening. This frees up recruiters to focus their energy on assessment and selling.
- Learning: The skills and capabilities we need at work are constantly changing. We now have vendor software that can intelligently recommend videos or learning programs based on your job role, experience and peers. There are systems that automatically read documentation and create micro-learning programs, and even systems that read and interpret an employee's writing or activities to recommend learning options.
- Leadership: Today, there are vendors that can intelligently assess characteristics of high-performing leaders by sending messages to employees to assess a leader's capabilities. Leaders can then receive personalized coaching and can view dashboards comparing their management effectiveness against their peers.
- Video interviews: There are a lot of companies, about 40% to 50%, who now use video-based interviewing. These videos go to a database and can be analyzed by AI to help determine an interviewee's mood; whether or not the candidate is telling the truth; and different things about the candidate's language, skill and education level.
- Sourcing: There are a lot of new tools available to help HR executives find candidates on the internet, a process known as sourcing. Sourcing is a manual, handcrafted process that companies undertake to find candidates for roles that are otherwise hard to fill, such as executives, salespeople and engineers. Now there is software that can search and intelligently match prospective candidates against the characteristics of a particular company and communicate with them to see if they're interested in the position.
- Career management: There are new tools that will intelligently look at patterns of career mobility inside of a company and point employees to roles that would be suitable for their skills.
- Sentiment analysis: AI in sentiment analysis is another big game-changer. Companies have a huge amount of survey data and feedback data, and AI can now quickly identify trends in the sentiment of employees. You can even locate geographies, offices or teams that are under stress by monitoring the sentiment of their emails and conversations. There will be more developments on that front; this is only the beginning.
Josh Bersinfounder and principal, Bersin by Deloitte Consulting
The heart of AI in HR is better-informed people decisions. Decisions such as: Who do I hire? Who do I promote? Who's in compliance? What's the chance of fraud? What's the chance of sexual harassment? These are all people decisions, and they're often based on judgments and have some degree of bias. Over time, AI in HR will be able to use unbiased data to help hire, promote and pay people on a more fair and unbiased basis.
AI is not perfect, and it can also institutionalize a biased or incorrect set of assumptions. Many large social networking companies use human intervention to train their AI systems -- which often recommend ads, tag photos or flag postings -- so we should do the same.
While the phrase artificial intelligence may imply smarter decisions, AI systems need training and monitoring to work well. In our domain, they can potentially harm a candidate's chances, damage an employee's reputation or create an incorrect performance appraisal. We have to be vigilant.
Challenges around using AI in HR
Though promising, AI in HR can present a number of issues, the biggest being the quality of the data collected by companies.
HR data is not managed effectively by most companies, and it is often in a number of different places. For instance, companies typically have five to seven different systems of record; therefore, applying an algorithm against any one data set can give you misleading results. It is a major challenge for companies to get all the people-related data into one place so it can be analyzed.
The second challenge is policy. Explaining to employees how their data will be used and having policies to ensure that it's not misused remains a challenge. Most employees agree to their company having access to data when they join, and companies inform them that the data will be used for positive purposes. Sticking to those policies is key.
The third is that there is no part of HR that's black and white. There's almost never 100% agreement on who the best candidate for a job is or who should be promoted. Almost every decision in HR has some amount of judgment. So a lot of these algorithms may be helpful to eliminate that human judgment or bias, but that does not mean they are 100% correct.