Now that the dust has settled, it’s time to take a critical look at AI. Recent developments in AI have transformed the market, with software vendors integrating third-party AI systems into their solutions. At SmartDreamers, we have done the same, integrating OpenAI at various stages of the candidate journey. Our tests showed real improvements in candidate experience, but they also revealed significant limitations that are often overshadowed by the current AI hype.
First, it’s important to distinguish between true AI and simple automation. Many vendors use rule-based algorithms, like chatbots, and label them as AI. While these tools automate processes, they are not powered by machine learning or natural language processing and thus shouldn’t be confused with AI capabilities. Developing true AI from scratch requires vast resources beyond the reach of most organizations.
Data Dependency
One of AI’s biggest limitations is its reliance on data. For AI to be effective in talent acquisition, it requires extensive datasets—data on job seekers, hiring patterns, skill trends, and much more. Many small and medium-sized companies lack access to this type of data, which limits AI’s ability to make accurate predictions or recommendations. Moreover, even with large datasets, if the data is biased or incomplete, AI can perpetuate or even amplify these biases, leading to ethical concerns such as discrimination in hiring practices.
Another challenge with data in talent acquisition is its scattered nature, making it difficult to centralize and train AI engines with data that represents the entire hiring funnel. Due to this, AI deployment is typically done in stages - focusing on specific tasks like resume screening or interview scheduling - rather than offering comprehensive oversight across the entire recruitment ecosystem.
Challenges with Keyword Matching
AI-powered resume screening tools often rely on keyword matching, which can be problematic. Candidates with non-traditional resumes or those who describe their experience differently might be overlooked if their keywords don’t exactly match those the AI is trained to prioritize. For example, a candidate may have the necessary skills but fail to use the precise terminology the AI expects, leading to potentially strong candidates being filtered out of the hiring process.
Response Times and User Experience
Another issue is the processing speed of AI systems. For example, when we use OpenAI in the backend to generate job descriptions from thousands of requisitions, we end up with response times that can take up to 30 seconds. While generating job descriptions directly in ChatGPT is faster, it has a significant drawback - it relies solely on external data, not your own repository of job descriptions. This delay may be acceptable for internal users like recruiters, but it’s less ideal for candidates. Long response times can increase drop-off rates, negatively impacting the candidate experience, particularly in competitive job markets where user experience is critical. Going back to our example with chatbots, while interacting with a chatbot can be impressive, it also often leads to longer times required to take an action, like finding out there are not relevant jobs in your area.
Lack of Human Judgment
AI lacks the ability to understand human nuances, which limits its application in critical decision-making processes like hiring. AI models like ChatGPT do not “think” - data is processed based on patterns and probabilities. While AI can automate certain tasks, such as resume parsing or job description generation, it cannot evaluate the complex qualities of candidates - like cultural fit, emotional intelligence, or adaptability—that are essential for many roles. For example, AI can’t assess a candidate’s interpersonal skills or potential for growth, areas where human judgment is indispensable.
AI and the Future of Talent Acquisition
While AI is becoming increasingly popular, we must remain cautious about its limitations. It can improve efficiency and automate repetitive tasks, but critical aspects of hiring - such as assessing soft skills, cultural fit, and long-term potential—still require human judgment. As we continue to innovate with AI, it’s important to understand its true capabilities and limitations and not be swayed by the AI hype or claims of breakthrough technology from vendors.