Generative AI in HR: ChatGPT, Hiring and Why Career Development is the New Talent Acquisition

GPT-3 – the artificial intelligence (AI) engine behind ChatGPT – is changing what we know about AI in HR, from talent management to talent acquisition.

After more than 30 years of research, Generative AI is a reality. Generative AI, like ChatGPT, is a type of AI that involves the use of deep-learning neural networks to create and generate content. We are witnessing the rapid development of Generative AI-powered tools and the evolution of technology, comparable to the introduction of the Internet or smartphones. 

The long-term implications of Generative AI in HR are hard to predict, but we can realistically predict the short to medium-term impacts of Generative AI on talent management and more specifically the impacts of Generative AI in talent acquisition.

An important thing to keep in mind is that Generative AI is still not ready for prime time. Like all technology, it will go through multiple setbacks before it becomes a reliable resource. But we can see where it is headed and would be remiss to not plan for Generative AI’s inclusion in HR. 

The Impact of Generative AI on Jobseekers

Generative AI in HR most immediately and largely affects the biggest group in the talent acquisition equation – jobseekers. There are two main areas where Generative AI will impact hiring for this group:

• Professional development
• Job application gaming

Professional development ensures a successful career, while job application gaming is, without a doubt, effective in landing a much-needed job right now. 

I’ll address how employers can adjust to this new reality. But first, let’s look at the most obvious examples of the new possibilities of Generative AI in HR.

Professional Development

1. Career advice. This new generation of AI could be instrumental in helping folks who are just joining the workforce by learning about their education, preferences, and personal qualities to provide career direction in a field of employment or even specific companies. Moreover, these machines could also assist with providing advice for specific situations in a workplace or career negotiations.

2. Identifying skill gaps. Generative AI can use information about one’s educational and professional background, as well as current skill set, along with the desired next step in their career to provide a jobseeker with a comprehensive development plan. Including a suggested curriculum, skills to learn, and references to educational content.

3. Upskilling. AI engines like ChatGPT are incredible learning resources. They are universal private tutors. This provides job seekers with the unique ability to not just passively consume educational content, but also ask questions and engage in a professional interaction about the content.

Job Application Gaming

1. Resume. The death of the resume was predicted for years, but it’s still here and is still required for the vast majority of job applications. Companies still manually or automatically use them to create an impression about an applicant’s fitness for a job. To a large degree, a resume is judged based on how close it is to the position requirements. 

How can Generative AI help a candidate clinch a role? Well, a candidate could now take a job description and ask the AI to adjust their resume to match it to the position’s requirements. Not only would the resume look much more relevant, but it would also be written with impeccable grammar and punctuation.

In the long term, this new-found ability for applicants to give an employer just what they want to hear will do much more to retire the resume as a hiring tool than anything in the past. The resume, as we know it, is probably, finally, coming to an end. But in the meantime, resourceful candidates will take advantage of the AI help.

Should we be upset about this? No. Employers will still need to know candidates’ employment and education history, but it doesn’t have to come from a resume. Resume-based screening has long been a source of hiring biases, while its ability to predict future job success was always incredibly low.

2. Skill and cognitive testing. Online skills and cognitive tests have always had the risk of being cheated by applicants finding answers online or asking somebody else to take the test for them. What we have with the arrival of Generative AI is a combination of both. With Generative AI, a candidate can get the answers immediately and in a succinct format that allows them to respond perfectly to the majority of knowledge, calculation, and logic questions within the time constraints imposed by a testing tool.

For employers that put a lot of stock into skill and cognitive testing, the availability of Generative AI to applicants is not great news. In short, these tests are now on the verge of becoming even less valid because gaming them just became a whole lot easier.

Is there anything to be done about this? It depends on the type of testing. Some of these tests should be done away with because, in all likelihood, they create more hurdles and discrimination in the application process than benefits to hiring quality. Other tests will have to be replaced by tasks that are not easily gamed. That includes various interactive assessments and personality-based assessments that I will discuss further.

3. Personality and behavioral testing. Unlike hard skills and cognitive tests, this type of assessment typically doesn’t have clear right and wrong answers. At least, not without context. But, Generative AI can be smart enough to understand this context and provide much more meaningful assistance to achieve better results in this type of testing. The critical question here is where the context for personality requirements comes from. A personality assessment based on a boilerplate competency model will be easily “cracked” by Generative AI just by knowing which role a candidate is being assessed for and the assessment type or vendor. 

The more advanced and the less obvious personality tests are, the harder they are to game. For example, a personality assessment that is generated based on some kind of job description analysis only requires providing this job description to the AI. Then from it, the machine will be able to infer which personality qualities an employer is seeking. The hardest behavioral tests to game are personality models created by Narrow AI (the likes of which are the focus of Cangrade’s work). These Narrow AI models are created not based on textual processing but by learning from specific employee success KPIs. The result of this learning is usually much more complex and impossible to guess by just reading a job description. Plus, there are other advantages to using a Narrow AI approach, including high predictive accuracy and low likelihood of biases.

The Impact of Generative AI in HR

Let’s see now which new talent acquisition opportunities Generative AI technology has in HR. I’ll examine four different areas, three that are traditionally seen as part of talent acquisition and one that usually belongs to a different HR department:

• Sourcing and Candidate Attraction
• Screening and Assessments 
• Interviewing
• Career Development

There is no question that Generative AI in HR will permeate each of these areas, but we have to evaluate where it will create efficiencies, where it will generate a qualitative improvement, and where it may turn into a source of systemic biases. 

Sourcing and Candidate Attraction 

1. Chatbots. Chatbots are already offered by many career sites today, but they are limited to a few topics like available job openings, job application processes, and general info about the company. This is about to change. Generative AI will turn career sites into an engaging and interactive experience, akin only to being helped by a human recruiter with unlimited time for answering candidates’ questions. These new-generation chatbots will not only help engage more candidates and compel them to apply but will also better qualify them and prepare them for the job application process.

We will see a lot of exciting updates in this area from software vendors in the upcoming months and years.

2. Job descriptions. Unlike chatbots, this is something that recruitment professionals can already do using unspecialized Generative AI tools, without waiting on HR software vendors.

Job descriptions that are used to advertise employment opportunities to potential candidates are a significant factor in applicant sourcing. Yet, most job postings are bland, indistinguishable from others, and targeted to no particular audience. They are usually produced by a bureaucratic corporate mill and used by recruiters “as is.”

Generative AI can take an unremarkable blurb, along with any additional company and role information, and turn it into a much more compelling narrative. Moreover, it can easily be targeted at different audiences relevant to the sourcing venues. For example, a job post could be targeted to appeal to Gen Z on a campus job board posting.

3. Review and validation. It’s important to note that Generative AI is not expected to be flawless any time soon and everything that it suggests requires a level of review and validation.

Screening and Assessments 

This is the part of talent acquisition where Generative AI in HR needs to be approached with utmost caution. This has to do with the very nature of this technology. Generative AI learns from a vast amount of sources, all of which (at least for now) are originated by humans. If we know one thing about humans, it is that they are a major source of prejudices and biases about other humans. Any AI that learns directly from humans will inevitably incorporate these biases into its results. AI vendors need to work extremely hard to control for these biases and be able to demonstrate why their no-bias guarantee can be trusted.

Let’s look at some examples.

1. Resume screening. There are many products on the market today aimed at analyzing resume data and turning it into meaningful information about a candidate’s job fit. A decade ago, this technology was already able to parse resumes and turn them into the well-structured data required for identifying educational backgrounds, skill sets, years of experience, etc. Generative AI does not have a lot to contribute to improving this area.

Some might hope that Generative AI could make more far-reaching conclusions about candidates based on how a resume is written and other less obvious aspects of it. And you can count on HR software vendors to explore this option. However, there are multiple reasons why the risks here outweigh the benefits. The major ones are:

1. Language with all its influences, dialects, and cultural backgrounds is an unreliable source for conclusions on anything outside the basic facts that it tries to convey (which are already captured by good old resume parsing).
2. The use of Generative AI in writing resumes will leave very little personality in a resume and a lot of personalization: to a position, to a company, and to a recruiter.

Ironically, the net result of using Generative AI in HR for resume screening will most likely be a preference for resumes customized by other Generative AI.

2. Skill testing. It might seem like Generative AI has everything needed for probing a candidate on a specific area of expertise. It’s easy to imagine it as a chat-like experience, similar to a technical interview. This chatbot may ask a candidate to write a sample of a customer communication or program code or ask you to demonstrate how you do something in Excel.

It’s a compelling picture, but it doesn’t pass the reality check. In reality, any such skill testing will inevitably turn into a “battle of AIs.” The main thing that it will test is not the skill in question, but a candidate’s ability to utilize other Generative AI tools to complete such tasks.

Likewise, aptitude testing is unlikely to benefit from Generative AI. It will bring the risk of systematic biases. Some of the existing aptitude testing tools, in particular those that are not relying on any type of knowledge, will continue to deliver better results than anything that Generative AI has to offer.

3. Personality testing. There most certainly will be attempts to position Generative AI in HR as technology that can help establish the personality fit of a candidate for a role. It seems viable that some aspects of this type of AI could contribute to this task. But this is also an area highly prone to bias and it needs to be approached very carefully and diligently. This topic deserves its own article, but here are the main points:

1. Generative AI is not known to be a reliable source of candidate personality data. It will require extensive independent studies to show that it comes close to existing tools (such as psychometric instruments).
2. Generative AI can play a role in deciding which personal qualities to prioritize based on sources like job descriptions, employee reviews, etc. But even then, the best we can hope for is that this technology will help recruiters augment the processes of personality success modeling, not replace it.

In the field of personality screening, the best tool, for now, will continue to be Narrow AI technology (one of the main areas of focus for Cangrade) that learns from specific quantifiable KPIs rather than qualitative inputs.

4. AI detection. Ironically, this might be the most needed contribution of AI in the area of candidate screening – detecting if a candidate’s response is nothing more than a machine.

Interviewing

For years, HR tech vendors tried to utilize AI for ranking interview responses. We see less excitement about this now than ever before because of increased privacy awareness and discrimination issues that came to the surface as a result of the algorithmic ranking of interview responses. Can Generative AI make it better?

An entire interview process conducted by a bot continues to be sci-fi. But some aspects of a job interview can benefit from AI help, and recruiters don’t have to wait to do it.

1. Interview questions. By providing the necessary context to a generative AI tool (like role, important competencies, and seniority of candidates) interviewers will be able to get a lot of useful ideas about what to ask their candidates in an interview.

2. Post-interview follow-up. Another area where AI can be a good source of ideas about how to handle responses to different post-interview situations.

3. Offer negotiations. Yet another area where these AI could help in HR communications.

The job of an interview is to establish a human-to-human connection. A good recruitment workflow should be geared towards utilizing automated accurate and unbiased tools for the early vetting of candidates and reserving the interview step only for those who were identified as a good fit through this process.

Career Development

Career development has taken on a new level of urgency due to ongoing candidate shortages, skill gaps, and employee turnover. Traditionally, career development is seen as part of post-hire talent management, not as part of talent acquisition. This is likely to change in the foreseeable future. Likely because of Generative AI.

There is no doubt that Generative AI in HR has the highest potential to supercharge learning and development. Some of this learning will be training on how to use Generative AI to be a more productive employee and work smarter rather than harder. But here I will focus on how Generative AI impacts talent acquisition.

The idea of hiring candidates that lack the necessary skills on day one and training them on the job has been unappealing to companies. The main reasons for this are time-to-productivity and training expenses. We are seeing a number of factors that will compel companies to reconsider this reluctance:

• Skill gaps
• Candidate shortages
• Reduced training costs due to Generative AI

In other words: there aren’t enough candidates that fit the strict skills and experience criteria required for many jobs today. But, there will soon be tools that allow you to upskill new hires to the necessary levels quickly and inexpensively. Moreover, Generative AI in HR will continue to assist these employees with skills that they need throughout their work.

This career development will need to start before hiring. When a new employee comes on board, they should already have a clear path to acquiring the necessary skills and a broader picture of possible career trajectories within the organization.

Conclusions

In summary, here are the main revolutionary impacts of Generative AI in HR from talent acquisition to talent management:

• Attracting and engaging candidates will be vastly improved.
• Screening candidates on specific skills will become even more challenging, but luckily less necessary.
• Screening candidates on cultural and personal compatibility for jobs will become even more important, and Narrow AI will still be the tool of choice for achieving that.
• Embracing learning and development during the recruitment process will be required instead of deferring it to post-hire.
• Onboarding and ramping up employees will become highly personalized with AI-assisted professional development.

Learn how Cangrade’s AI for HR can help you hire the right candidates, efficiently and bias-free. Request a demo.