What you should consider during choosing an AI software
It is important for recruiters, recruitment team leaders, HR Directors and all those who make recruitment decisions and what the recruitment process will be like in the organization to understand the specifics, not only the benefits but also the risks of any automated recruitment tools they plan to implement in their practice.
Artificial intelligence-based tools will be a hit in the coming years in the field of recruitment and recruitment automation. As we described in the article “Artificial intelligence in recruitment” AI technologies have significant potential to optimize or even fully automate many routine and voluminous recruitment tasks and already demonstrate impressive results. At the same time, they have some limitations and carry certain risks that can affect the quality of your selection, the work of the organization and candidates.
Risks of using AI tools
We consider machines by default to be impartial and fact-oriented, not personal judgments but in fact they indeed are. The problem is that the” facts ” on the basis of which they generate their solutions are supplied to machines by people. How does it happen?
An important aspect of artificial intelligence is that it is able to learn. “Machine learning” allows AI to take a set of data, for example, a number of decisions made by man and on their basis to create an algorithm how to make similar decisions for future pieces of information.
Here lies the main risk: if past decisions that AI has studied have contained human, cognitive and general biases so the AI will include them in its decisions as well.
This means that artificial intelligence is critically dependent on the information supplied to it for studying not only its volume (AI requires large amounts of data, for example, from several hundred to several thousand CVs for a certain position) but also the quality, as it will directly affect the quality of subsequent decisions.
AI does not substitute for human
The previous problem has another important aspect. We can teach the AI to focus on truly forward-looking (having predictive power) indicators of capacity, ignoring unnecessary or spurious signals and factors. In this case, the AI will act much faster and more efficiently than recruiting people.
But faster or cheaper forecasting does not solve the fundamental recruitment problem: the need to have reliable criteria or indicators of the employee’s success in his or her position or in the company. Only having reliable employee performance indicators you can create meaningful models to predict future performance and quantify a person’s suitability for a role or job.
That means artificial intelligence does not replace but alters the work of recruiters; the success of using AI algorithms in your company will primarily depend on your ability to determine significant indicators for you and teach them AI. This requires human ability (in addition, recruiters are still needed when it comes to building relationships, improving employer brand, negotiating and all other activities where a “human touch”is needed). If your recruiting is now based on intuitive decisions, chaotic, depends on personal preferences, does not use the confirmed criteria and indicators the tools based on AI will not improve it (rather even worsen).
Not all AI is really AI
Although many of the tools offered to recruiters are advertised as AI, AI-enabled programs or “AI assisted” are not always true.
Some tech companies that promote their AI tools don’t even use machine learning but rely on complex decision trees to move around depending on answers and anticipate candidate questions. These types of technologies are not artificial intelligence. They go through the options (a large number of options) but do not learn. They may be useful but they are not flexible or “smart.”
At the moment, it’s a marketing buzz that could potentially distract from more important issues: the company has a lot of data with which it does nothing because it does not know what they can give or what to do with them. Or the company doesn’t even collect data about the people it actually has yet. Or the company does not have a clear understanding of how to allocate from the existing data really significant for hiring indicators. Machines will not solve these problems.
Like any technological innovation AI technologies raise a number of ethical questions the answers to which are yet to be found.
Artificial intelligence can improve our ability to match people to the right jobs and that is a definite advantage. However, the use of AI can lower ethical standards for evidence-based selection methods (e.g., well-designed structured interviews, valid psychometric assessments and job modeling).
It is important to understand the ethical implications of using AI for employee recruitment and selection. Among them:
- data collection about candidates (there is a difference between what we can and should know about people and this difference is a matter of ethics) and the protection of the data;
- providing equal and fair opportunities to all candidates;
- providing feedback on the selection results with the help of AI and the potential impact it may have on the candidate;
and so on.
Recommendations during choosing technological solutions using artificial intelligence
Understand how the Software you are going to buy works
On the basis of what principles and models artificial intelligence analyzes data, learns, determines significant patterns, makes decisions.
Be very careful with the data you transmit to the software
Remember that their quality and volume are critically important to future results.
Make sure the developer company is aware of the bias issues and is working on it
Removing human cognitive distortion is one of the most important advantages of implementing AI algorithms.
Rate potential ethical risks
Identify the main challenges in the area of personal data collection, personal data protection, potential impact on the reputation and brand of the employer that your company may face in the future during using new solutions.
Of course, set up your recruiting to have a good understanding of exactly how new tools should complement and improve the work of your recruiters.