Lots of HR software vendors have released - or will soon release - product enhancements that add AI capabilities to their products. Some of these are definitely cool and value-adding. But some will be novel, faddish things that won’t have great staying power and some will carry price-tags that are not commensurate with the value customers might eventually receive.
Let’s look at one of these.
Job description generation in the AI world
Here's the AI promise. Vendors believe that Recruiters would love a tool that takes a lot of the tedium out of writing job descriptions. Moreover, vendors are betting that AI tools can be trained on thousands of job descriptions that other firms have used. That training might help the AI tool suggest additional skills that should be added to the recruiter’s initial effort. All in all, AI tools might create a more complete, better written job description in seconds. That sounds great – or is it?
There's a major problem. AI tools that read other firms’ job descriptions will write a job description for you that reads like a job description from any number of your firm’s competitors. While it might be more complete, it likely won’t contain anything that makes your firm’s uniquenesses stand out.
Chapter Seven of Lars Schmidt’s book Redefining HR is worth a serious read. Please check out this excerpt:
Let’s be honest. Most job descriptions suck. The reason they suck is, in a sense, archaeological. The last job posting sucked. And the one before that. And the one before that. Job descriptions are often compliance-driven recycled relics dusted off from the last time that the position was filled.
The reality is that job descriptions are actually one of the most crucial employer brand assets you have. They often have the most traffic. They help inform and align prospects to your roles. They don’t have to suck. They can sing! Okay, that’s hyperbole – but my point is that even incremental advancement in job descriptions can feel monumental as the bar is so low.
There’s a mountain of gold in that excerpt. What a savvy reader will note is that:
- The job description generator tools out there are being trained on other firm’s job descriptions that likely ‘suck’ too. This is a classic GIGO (i.e., garbage in, garbage out) problem with these AI tools potentially re-purposing substandard, generic or uninspiring content. You can’t expect greatness with this stuff for input.
- The compliance driven comment is telling. Many job descriptions cannot excite a jobseeker. The emotion, employer branding, culture, fun, etc. has been scrubbed out of most job descriptions. An AI tool that has ingested thousands of these soulless pieces of prose will only generate more job descriptions that suck.
- Job descriptions are indeed a mirror to a firm’s culture. They will either reflect the culture of the company or the unimaginative aftereffects of a recruiter or operational manager. Rarely are job descriptions written for their intended audience. They’re written to appease Legal, HR, Operations and other managers who aren’t always the go-to people for writing brilliant copy that inspires, motivates and, equally important, paints a realistic view of the work ahead for applicants. AI tools will reflect the blandness of thousands of aggregated job descriptions and none of what makes your firm or this role exciting, compelling and competitively interesting to job seekers.
In the war for talent, the creation of long, dry, generic and undifferentiated job descriptions is a non-starter. An AI-tool that creates these is of low value. Furthermore, recruiters who use such tools would be wise to only accept a modest percentage of the suggested copy and supplement that with material from their own firm that excites, motivates and energizes jobseekers. Use the space to get the cultural alignment your firm needs and triggers the people you really want to apply to complete the process.
Finally, ask yourself - are our job descriptions generating responses from the best and brightest candidates we really want to attract or are they returning too many so-so candidates (or not enough candidates)? If the AI-generated/AI-enhanced job descriptions read the same way as your current job descriptions, then why would you think they’ll deliver the quantity and quality of jobseekers you really want? The strategic goal should be to Win the War for Talent not write more job descriptions quicker, longer and cheaper. Get the strategic question right folks to really understand what this capability is really worth!
What should this AI capability cost?
The ability to generate copy that is the result of ‘training’ from thousands of similar work products is a straightforward AI task. The cost, if you will, is in the data processing power needed to do the initial training and in operating the data center that hosts the Job Description Generation program. Those incremental costs are likely quite low.
Logical software buyers want to look at the costs/benefits of a new technology. In their ROI analysis, they’ll want to know if the new tech will reduce the labor required to do a task, power more efficient processes, reduce errors, and/or, most importantly, deliver a superior business result. A job description generator might help recruiters craft job descriptions more quickly but are the outputs helping them hire more, better people? I have my doubts.
You might also doubt the ROI claims as the AI-tool could hallucinate some job description content. You’ll need a recruiter to review, carefully, the suggested copy to ensure it meets your firm’s standards. Then that recruiter will need to alter that copy so that it better reflects the culture of your firm and also describes, realistically, what the role will entail. That’s a lot of labor that the AI-tool isn’t removing and that labor can adversely affect any potential ROI.
Today, you can find free job description generators online. Workable has a pretty good one and it’s free! I tried it out and it does a solid job quickly. It’s free price point is something any HRMS vendor should note, especially if they are thinking of charging for their job description generator.
Should Talent Acquisition or HRMS vendors charge a premium for this AI-powered capability? No! It’s a very small piece of functionality that doesn’t really deliver game-changing or competitive advantage for its users. It’s really a nice-to-have functional add-on and not much else. In fact, this very incremental enhancement to a Recruiting module is exactly the kind of thing you’d expect a vendor to deliver as part of their normal product enhancement processes – and those don’t cost extra.
Let’s remember that SaaS (Software-as-a-Service) vendors - or any vendor that sells via a subscription - must re-earn the right to this subscription each and every month. Vendors should have the mindset of always delighting (not gouging) customers month after month. That means that vendors are supposed to continue innovating and that innovation is baked into the pre-existing subscription price. Yes, a material innovation (e.g., an all-new module) may warrant an additional fee but incremental enhancements (like many generative-AI add-ons) should be part of the service.
Look past the novelty
Not every piece of software will be long-lasting or add great value. They all have a “Best When Used By” date on them if you look for it.
If you need a generative AI tool to deliver a highly personalized output (e.g., resumes, job descriptions, etc.) then you might be disappointed. Where generative AI can shine is in the opposite vein where a dispassionate, boilerplate response is acceptable (e.g., a chatbot that answers questions about employee benefits enrollment). Don’t pay for a generative AI tool that can’t deliver the highly personalized and focused outputs you need and don’t fall for the hype and high prices for something that may not be worth the price.
Additionally, be prepared to cease using some tools after the novelty has worn off. In the job description generator example, users should note that job seekers will use differ AI tools that will cause your Recruiting inbox to be flooded with thousands of lookalike, undifferentiated resumes (see this piece on how AI kills resumes). An AI tool that you thought would deliver years of value may turn out to be far less useful in a few months. Don’t believe me? Read this excellent piece in Wired magazine as it details what happens when a jobseeker’s AI bot completed 5,000 job applications for the jobseeker and how recruiters hate this technology. Bottom line: Why pay for something for decades that only provided a small blip of value short-term? Don’t let a short-term functional novelty’s cost become a permanent addition to your subscription price.
Software buyers may need to educate vendors as to what AI capabilities are really solving strategic problems and will be long-lived. That means customers must tell vendors “No!” to some of these proposed up-charges and remind vendors that they have to re-earn the customer’s business every month.
We all need to remember that AI-capabilities can run the gamut from highly useful to ho-hum to counterproductive. It’s worth noting that just because a vendor can create something doesn’t always mean they should. Some of these AI add-ons will undoubtedly be short-lived or of marginal value. If vendors can’t figure this out, you will need to do it for them.