AI is everywhere, and it should be no surprise that many candidates are using it when applying for jobs or to help answer interview questions. The problem for hiring managers is that this makes it even harder to sort through applicants to find the right fit, especially in tech startups that have been focused on finding experienced talent with technical or specific sales expertise.
In this blog, we will go over some of the top signs that a candidate may be relying on AI-generated answers when applying to or interviewing for an open go-to-market (GTM) position:
Gen AI in Recruiting
The current wave of artificial intelligence ushered in by popular solutions like Open AI’s ChatGPT has been compared to a Pandora’s Box for technology, bringing a constant stream of new innovations and uses for Generative (or Gen, for short) AI seemingly every few months. Like the days of Apple’s “There’s an App for That,” new Gen AI-powered solutions for many different industries and use cases are emerging almost every day.
The wide availability and functionality of these tools often also mean that virtually everybody has access to some sort of LLM (Large Language Model) to generate text, images, coding and even video. Recruiting is naturally an area that is seeing a lot of artificial intelligence use, from hiring managers leveraging it to automate talent acquisition activities at scale while many job seekers turn to it to help with content outputs such as their CV.
Why (and When) Does it Matter When Hiring?
There are a few downsides and pitfalls when relying too much on current generative artificial intelligence toolsets, particularly for content that requires accuracy. User error can often exacerbate these issues. LLMs can be prone to “hallucinating,” or essentially making up details and passing off this incorrect information as fact. If you do not double check the output – or are unfamiliar with the actual facts of the topic – and share it unedited, chances are that you will be presenting false information somewhere in your content.
Another unfortunate double-edged sword with using Gen AI lies on the behavioral side, where too often we are prone to copy and paste what the LLM gives us, hallucinations and all. Even if a candidate genuinely just wanted to touch up the writing quality in their resume, relying on artificial intelligence may leave them with a CV that grossly misrepresents their job history and experience.
How to Spot AI-Generated Applications and Answers
The good news is that there are a few telltale red flags to look for if you suspect a candidate is abusing AI to sound better for the job they are applying for. LLMs themselves are not exactly thinking tools – as the name suggests, they process language first and foremost, and often what you get out of them is a reinterpretation of content that already exists. Due to the prevalence of hallucinations, a rehashed output also has a chance to be incorrect if the input is not carefully constructed to avoid these by the user.
These and other factors create common signs of artificial intelligence involvement, and keeping an eye out for these can help you determine if candidates are relying on AI tools for applications or interview screenings.
The Language
While not always the most surefire method, looking at the language used by the candidate can help determine if they are using AI-generated content to try to sound better. There are a few signs to watch out for:
- Repetitiously using corporate-sounding buzzwords like “savvy,” “bespoke,” “seamless,” “comprehensive,” “robust,” “alignment” and “inclusive”
- Lots of generic phrases and terms that seem like they may be acting as placeholders for actual information
- Tone and point-of-view inconsistencies when referring to subject nouns or actual events
- Awkward use of more conversational elements, especially when switching with formal language
- Lack of personal anecdotes or personalization of the narrative being told
However, there are a few reasons why this approach is not always clear-cut for sniffing out AI reliance. One of the biggest is that quite a few of these are common faux pas made by real people, even experts who may not be the best at communicating topics they are less familiar with. The trick is applying a reliable benchmark to help – repeating the same three buzzwords, avoiding providing real examples and showing a consistent lack of understanding of their supposed field are a few you can fall back on to gauge a candidate.
Too Good to be True
Before ChatGPT was even conceptualized, a philosophy professor summed up one of the biggest idiosyncrasies in AI with the Paperclip Maximizer thought experiment. Without the human contextual decision-making we are so used to, artificial intelligence focuses on what it is told to do, even if it is at the expense of everything else in existence if no guardrails are in place.
Less dramatically, common Gen AI solutions may take a prompt at face value – which means that if a job seeker asks one to simply improve their resume, it will probably take this to the logical conclusion. If an entry-level SDR candidate comes in with executive-level achievements on their CV, then there may be a good chance that they copied an output from a LLM after asking it to make them sound like a better sales rep.
Where Have I Heard This Before?
If you read any article published 2022 that starts with “in today’s world…” or “in the fast-paced world of…,” chances are it was at least partly AI-generated. LLMs probably have trillions upon trillions of examples of human-generated language, but (depending on the model) they seem to most often default to those they see the most, deem most popular or some other arbitrary metric.
This has left a lot of those unfamiliar with AI hallucinations sharing content that sounds almost exactly the same as content that already exists somewhere else. If your applicants turn in resumes that sound just like the job description, they probably ran it through artificial intelligence.
Checking Their (Not) Work
The most obvious signs of using AI-generated content will appear when a candidate fails to edit and personalize their submission, including the blatant examples like placeholder text and metacommentary from the LLM directly as well as things they may have missed. While OpenAI has waffled back and forth on adding distinct watermarks to ChatGPT’s outputs, some elements like formatting or SearchGPT UTMs will still sneak through in unedited work.
Often, the most telltale sign will be a mix of different red flags as if someone is copying and pasting directly from artificial intelligence, they probably missed any hallucinations already. Chances are also that a candidate with this poor level of attention to detail will out themselves as a wrong fit before you even have to do a deeper review of their submission.
Making the Right Hire When AI is Involved
AI involvement in talent acquisition is not necessarily a bad thing, and in fact many tech startups are increasingly seeking go-to-market (GTM) candidates that have experience with artificial intelligence tools already. However, the key challenge is in finding candidates who leverage these tools to augment their expertise, not replace their labor entirely.
It is also important to note that despite claims otherwise, tools for detecting AI in writing or other formats will not be able to catch it 100% of the time (if at all). Human intuition – and experience – is still required to make the call on when artificial intelligence is involved, with the burden being left to recruiters for determining if a candidate is the right fit or not for your goals.
Source Your Right Fit Candidates Faster with Betts Connect
Finding candidates who showcase their authentic selves can be challenging in the digital age. This is where Betts Connect comes in – our hiring platform connects you with pre-vetted sales, marketing and customer success professionals who have been thoroughly screened for their genuine experience and skill sets.
Contact Betts here to learn how Connect can help you identify and connect with experienced GTM talent in tech who will help you scale your revenue up and grow your organization.