This monthly feature is where I bullet the latest developments in using generative artificial intelligence to help you in your career (and life). This first one comes courtesy of Ethan Mollick, author of the “One Useful Thing” Substack, with commentary from me about how it might apply to our field – the goal of the note below is for you to learn the basics of how to describe the various differences between AI tools (with a hot tip for you to start talking to AI rather than writing to it):
- What is a parameter model? Parameter numbers refer to the complexity of the AI model (model meaning the same as LLM, which is the universe of data from which that particular AI tool draws). Larger numbers mean the models are more capable but are resource-intensive.
You can run the smallest models on your phone or laptop without modification, but specialized hardware is required to run the larger parameter models. The smaller models aren’t as powerful but make up for it in other ways. - What can be the benefits of smaller models? In addition to being able to run more easily, smaller models are narrowly focused and can do tasks well in a niche area. Generalist models are broader.
So for assistance with our line of work, you likely will use smaller models that dig deep into that niche, perhaps preparing Section 16 forms, for example. - The philosophic difference between copilots and agents. Copilots are smaller models designed for limited tasks and tend to be more predictable with fewer hallucinations. In comparison, agents are larger models that lean into the full power of the LLM.
- Differences in experience with copilots and agents. Copilots – which, as an example, are often a feature included in your phone – tend to do better answering straightforward questions with direct responses. In comparison, agent voice chats engage in natural conversations, with interruptions and fast flow – and even subtle tonal changes. It feels very human (e.g., fake breathing, pacing and intonation).
Your Hot Tip: Voice capabilities are going to change how most people interact with AI systems. Voice and visual interactions are more natural than text and will have broader appeal to a wider audience. The future will involve talking to AI, so you may want to start using voice for AI to get a feel for how it works.
Authored by
Broc Romanek