A recruiter’s thoughts on AI: Even skeptics (me) can’t ignore it

By Anastasiia Kuzmenko, head of talent acquisition at F1V

I'm quite skeptical about AI's current role in recruitment. While it already can automate some tasks, it’s not ready to fully replace human judgment. Only 30% of companies use AI for hiring, according to a recent report by recruiting platform Arya. And I guess I know why.

AI tools for HR and recruiting are popping up like mushrooms — there are hundreds. Still, there isn’t one tool that properly covers most of my team’s needs, including candidate search, automated sending of emails and LinkedIn messages, pulling contacts, and syncing everything with our CRM.

In a few years, however, this may change. Several reports I’ve seen recently say roughly 80% of businesses expect to start hiring with AI soon.

This is my essay on this topic.

Test and toss aside

Completely ignoring AI today is like driving a cart on a highway.

With this in mind, I decided to start by testing AI tools for everyday operations. My goal is to find at least 3-4. Yes, most won’t stick around, but you have to keep looking at what tools people are using right now. It must be a constant process in any modern company.

To make this testing fun, I try to gamify it in my department. Everyone spends a month experimenting with a new tool and then shares their impressions during Monday calls.

Where AI helps us

AI can quickly find and analyze data (ChatGPT, Gemini by Google, Claud, Perplexity). Researching a new job market now takes about an hour instead of an entire day.

Ask AI about hiring specifics, salaries, and taxes in a particular country. Add to your prompt that you’d like to hear must-know insights about the market. You may get interesting answers.

With AI, we also automate note-taking during interviews in Ukrainian or English. This allows us to focus more on the conversation and observe candidates' body language and reactions. Our pick is tl;dv, which creates meeting notes and summarizes discussions based on the prompts we provide.

It now takes the same time as to make a cup of coffee to craft a candidate summary and send it to a hiring manager. Since recruiters on my team conduct about 15 interviews a week (sometimes even 10-12 a day), this is a huge help.

Other routines AI can streamline: writing job descriptions, test assignments, and rejection letters.

Our team also uses AI to find new ways to source potential hires. For instance, it helps us refine Boolean and X-ray search, identify “donor” companies for headhunting. Not bad, ha?

Applicant tracking systems (ATS) aren’t falling behind the AI trend either. I’m satisfied with how our ATS TalentLyft parses resumes and automatically organizes candidate info in a database. It also has a virtual recruiter assistant that helps write rejection emails or other texts you need.

Currently, we’re also testing pulling contacts via Contact out and then working with this data in the ATS.

Where AI can be better

AI gets worse when it comes to automated candidate sourcing, outreach, application screening, skill assessments or personality tests, interviewing… That’s pretty much the entire recruitment cycle.

There are AI platforms automating each stage, but I wouldn’t entrust everything to AI. Now it works better when assisting my team in small operations where a human touch isn’t crucial. Otherwise, the risk of losing a good candidate would outweigh the perspective of finding one.

It's a bit of an exaggeration, but I feel this way now about delegating tasks to AI: I wouldn’t want the technician for my plane or my doctor to be hired solely by AI.

We're waiting for more features and improvements in People GPT. This tool sources potential hires via job boards and social media, analyzes CVs, and allows users to send messages to the specialists they fish out.

Its filters are versatile, but search results are often irrelevant. The platform usually struggles to find enough professionals with specific backgrounds, such as C-level, B2B, marketing, or sales. That’s why People GPT can’t replace LinkedIn yet — we have to check everything on LinkedIn in the second go.

Let's not forget that candidates also use AI to create CVs. Many don’t even edit them, resulting in profiles that look alike and don’t accurately represent their true skills. This makes it even harder to find talent, especially if recruiters also screen resumes with AI.

I'd say AI is still not super clever to replace a human recruiter, but it's hardworking and can do a lot with clear instructions.

More AI recruiting tools

Here are some popular hiring AI tools I didn’t mention above that you can try with your team.

Sourcing candidates. We’re testing HeroHunt.ai (unlike many tools, it works with LATAM) and Linked Helper. You can also try recruitin.net, RecruitmentGeek, SeekOut, Entelo, Fetcher.

Sourcing contacts. We’re testing Snov.io. You can also try SignalHire (works with LATAM), Kendo, RocketReach, Hunter.io, Wandify.io, Finder Expert, FindThat Lead, SalesQL.

Data parsing. OctoHR, Entelo, Linked Helper, Crystal.

Creating job descriptions and visuals. Easy-Peasy.ai, Loom, Textio, HeyGen, Synthesia.io, Stable Diffusion, Midjourney.

Communications with candidates. We’re testing Text Blaze. You can also try Magical, Remail.ai, FlyMSG.

Hard skills assessment. Codility, HackerRank.

Interview analysis. Hirevue, Metaview, Hireflix, Easyhire, Videoask, Talocity, Fireflies, Turboscribe, Pymetrics, TimeOS.

Analytics. Gloat, Beamery, Burning Glass Technologies.

Chatbots for hiring. Olivia by Paradox, Mya Systems, XOR, Botsify, Drift.

Platforms for creating chatbots. Dialogflow by Google, Microsoft Bot Framework, ManyChat, TARS.

Also, check out the AI stack mapping with over 50 productivity tools we compiled earlier (at the bottom of this article).

The thoughts I just shared came to my mind after moderating a discussion at South Summit 2024. You can watch this panel on YouTube.

Subscribe to our LinkedIn newsletter to get one useful article every week via email and on LinkedIn. We promise each story to be insightful and take less than 5 minutes to read.

More posts like this: