How HR Tech vendors are (not) using Generative AI effectively

Throughout this year, the term "Generative AI" has reverberated across the technology and business sectors. As we've journeyed through the year, the potential of Generative AI has seized the attention of innovators, entrepreneurs, and industry leaders.

In the HR Tech space, Generative AI has been particularly impactful. Its potential to revolutionize traditional HR processes, from recruitment to employee engagement, has ignited considerable interest among HR professionals and tech enthusiasts. The capacity of Generative AI to 'invent' an infinite array of skills and provide highly specific, data-driven insights is viewed as a game-changer.

However, while Generative AI models offer immense potential in the HR Tech space, current implementations by vendors only scratch the surface of what's possible. It's crucial to discern what HR Tech vendors are truly offering through Generative AI and to distinguish marketing hype from actual effective functionality.

To assist you in understanding this, we've meticulously analyzed the Generative AI offerings of over 20 HR Tech vendors. We've examined the extent of Generative AI functionality embedded in vendor platforms, the underlying Large Language Models (LLMs) used, and the use of external versus internal data in generating insights.

Beamery was the first HR Tech vendor to launch their TalentGPT offering, which leverages its proprietary AI, as well as OpenAI’s GPT-4 and other leading LLMs. Its effectiveness lies in its seamless User Interface, which is integrated directly into the platform. However, for other vendors, the Generative AI functionality often seems like an afterthought and feels detached from the main platform.

Another notable theme is the actual use case for Generative AI. For most vendors, it serves as a text interface for interacting with the underlying platform. In some cases vendors have existing AI capabilities. These vendors generate insights using their existing AI models and use Generative AI to interact with these underlying models.

Regarding underlying LLMs, most vendors are using OpenAI’s GPT3.5 or GPT4. Centrical has announced the release of AI Microlearning, their integration of OpenAI’s generative AI, ChatGPT. Similarly, HiringThing has announced AI-Enabled Job Descriptions, which lets users leverage OpenAI’s GPT technology to write dynamic, professional job descriptions in minutes. Interestingly, no vendor mentioned the use of any open-source LLM.

A key insight from our analysis is the limited use of internal data in Generative AI applications. Most vendors are focused on recruiting or learning and specifically on generating job descriptions and content. There is limited contextualization from an organizational perspective, making the use cases slightly less relevant in the current era of hyper-personalization.

There appears to be a trade-off between speed-to-market and the actual value that the use case will offer to the organization. This is also why there are a limited number of vendors talking about training or fine-tuning LLMs on an organization’s data, as it requires a deeper understanding of the use case and the organization's data. The transition to fully leveraging this technology is complex, requiring new tools, systems, and skills, and the industry is still in the early stages of this transition.