The matrix reviews applications for this technology through the lens of both risk and demand. Marc notes that in the category with low risk but high demand for this technology (copy editing, learning, marketing, code reviews etc.), we will see new ways to work with and learn from it happen fast. Corporate Learning falls right behind marketing in the low-risk/high-demand quadrant, as both disciplines share the need to interpret consumer needs and behaviours and to communicate regularly and persuasively.
Marc believes that if Language Learning Models (LLMs) can be tailored to train on smaller pools of private corporate knowledge while retaining their conversational spark, much potential will be unlocked. Today, the answers can still be unhelpful and generic, even with expert instruction, and this is substantial because they are trained on generic, public-domain content. Customizable LLMs should, in theory, be able to suck up the specifics and detail of a corporate learning and knowledge corpus and answer employees’ questions with much greater context. Fine-tuning an LLM is possible today, but it’s too early to judge definitively that they will achieve the nuance without losing any other advantages.
Link to the article discussed in this podcast: https://hbr.org/2023/03/a-fram
Further articles by Marc Ramos:
- Learning 3.0: A data-fueled, equitable future for corporate learning https://www.chieflearningofficer.com/2022/08/29/learning-3-0-a-data-fueled-equitable-future-for-corporate-learning/The Nutrition of Training Content & Six Regenerative Learning Considerations https://www.linkedin.com/pulse/framework-picking-right-generative-ai-project-marc-steven-ramos/
- The Cornerstone Acquisition of EdCast: Three Essential Hopes from an Alternative Perspective: https://www.linkedin.com/pulse/cornerstone-acquisition-edcast-four-essential-hopes-from-ramos/?trk=pulse-article