ChatGPT Revolutionizes Recruitment at Emergent

⚡

Key Takeaways

1Mukund Jha from Emergent uses ChatGPT to evaluate candidates, aiming to eliminate human biases in recruitment.

2New York's Law 144 mandates bias audits for automated tools, while the European AI Act classifies recruitment as high risk.

3Mukund Jha's practices raise questions about the transparency and social impact of AI assessments.

💡Why it matters — The integration of AI in recruitment could transform HR practices, but it raises crucial ethical and legal issues.

Emergent and the Use of ChatGPT for More Objective Hiring

In the business world, Mukund Jha, head of the startup Emergent, has opted for an innovative approach to recruitment. He uses ChatGPT to analyze interview transcripts, aiming to complement and refine his personal judgment. This method seeks to reduce potential biases from recruiters by relying on algorithmic evaluation.

For each hire, Mukund Jha compiles a detailed dossier. This includes the full transcript of the interview, the prompt used to query ChatGPT, and the score generated by the AI. These elements are then compared to human evaluations. With a team of over 100 employees and the goal of hiring 30 additional engineers by the end of March, Jha aims to generalize this approach across all of Emergent's recruitment processes.

Emergent boasts annual recurring revenues of $100 million, just eight months after its launch. The startup has also raised $70 million in Series B funding, backed by investors such as Khosla Ventures, SoftBank Vision Fund 2, and Lightspeed.

The Impact of Legislation on the Use of AI in Recruitment

The use of AI in human resources is not an isolated phenomenon. According to the Society for Human Resource Management (SHRM), a quarter of companies had already integrated AI into their HR processes by 2024. In New York, law 144, effective since 2023, mandates annual audits to detect biases in automated tools used in employment. In Europe, the AI Act classifies recruitment as a high-risk activity, requiring special attention.

In this context, Mukund Jha compares the results obtained by ChatGPT with evaluations from his teams. Sometimes, the AI identifies details that human recruiters have missed, and sometimes it’s the other way around. This dual evaluation aims to ensure greater objectivity in the selection process.

The Challenges of Transparency and Data Retention

Mukund Jha ensures that a precise record of each recruitment is kept. Each dossier contains the interview transcript, the prompt sent to ChatGPT, and the score assigned by the AI. These records also serve to justify rejection decisions to candidates.

However, the issue of data retention arises. OpenAI deletes temporary conversations after 30 days in its public interface, but exchanges via API or Enterprise are subject to different rules. Candidates are not always aware of where their data ends up, and no legislation requires employers to inform them. OpenAI's terms of use state that users retain ownership of their data, although the company may use it differently when submitted via API.

The Role of Language in AI Evaluation

ChatGPT's evaluation relies on text analysis, without consideration for the candidate's appearance or physical behavior. However, the language used can reflect social or ethnic biases. When Mukund Jha asks ChatGPT to assess "clarity" or "intellectual honesty," the model draws on a vast database of text to formulate a response.

A candidate who expresses themselves directly may be disadvantaged compared to another who uses more elaborate language. Research conducted by academics from Cambridge shows that large language models can associate certain speaking styles with perceived skills, thereby introducing social and ethnic biases.

The Impact of AI on Company Culture

By sharing his method, Mukund Jha indirectly imposes a filter on candidates. Those who are uncomfortable with the idea of an AI evaluating their interview may not fit into Emergent's company culture. Other specialized tools, such as HireVue or Eightfold, already incorporate audits and provide explanations to candidates. Emergent, on the other hand, favors a more flexible and less formal approach.

Industry figures, such as Becky Frankiewicz from Manpower Group, emphasize that AI can process resumes more quickly and learn to avoid biases. Nathalie Scardino from Salesforce highlights the importance of human judgment in assessing candidates' learning abilities. Julie Sweet from Accenture has already integrated AI into internal promotion processes.

As long as legislation remains less stringent, AI will continue to play a role in hiring decisions, raising questions about its long-term impact on HR practices.