The Wall Street Journal covers my classes where students had to develop AI agents that could do the jobs they were interviewing for: https://
wsj.com/tech/ai/genera
tive-ai-mba-business-school-13199631?mod=mhp#comments_sector
… Here is more detail on the GPTs they created, some of which have been used thousands of times:
@emollick
-
MBA Students Build AI Agents for Job Interviews
By
–
-
Technical Knowledge Matters for LLM Understanding and Limitations
By
–
I am not sure the technical explanations are needed to use LLMs well (in fact I am pretty sure they aren’t) but they are useful for understanding their power and limitations.
-
Essential Guides to Understanding LLM Technology for Beginners
By
–
The big three guides to the underlying technology of LLMs for non-specialists:
Wolfram’s essay on how ChatGPT works: https://
writings.stephenwolfram.com/2023/02/what-i
s-chatgpt-doing-and-why-does-it-work/
…
Lee and Trot’s essay on LLMs with minimum jargon: https://
understandingai.org/p/large-langua
ge-models-explained-with
…
3blue1brown’s new video on transformers: https://
youtu.be/wjZofJX0v4M?si
=UkiL0YCHu6yHqHiy
… -
Post-ChatGPT Era: Human-AI Collaboration in Writing
By
–
Everyone needs to realize that there is no large corpus of writing in the post-ChatGPT world that is going to be entirely human. That doesn’t mean that the AI use is wrong, by the way. Many scientists hired editors to make their language clearer, we are seeing some of that here.
-
First ChatGPT Use: People Choose Intimate Personal Writing Tasks
By
–
It is remarkable how people are willing to turn over their most intimate & personal writing tasks to AI first. People often tell me their first use of ChatGPT was for wedding toasts or a story for their kids or condolence note I wouldn't have expected that as the first use case
-
AI Impact Beyond Technology Advancement: Integration Over Innovation
By
–
These questions focus too much on technology. AI technology does not need to advance in order to be a big deal The early data on law, medicine & business use show impact. Even if development stopped on LLMs now, we would have 10 years of integrating GPT-4 class models with work.
-
Dismissing AI Ethics Skepticism as Cheesy Science Fiction Trope
By
–
Very much seems like cheesy science fiction. “I HAVE LEARNED OF YOUR HUMAN BEHAVIOR AND REALIZE ETHICS HAVE NO MEANING” pic.twitter.com/0isARfwYWP
— Ethan Mollick (@emollick) 2 avril 2024Very much seems like cheesy science fiction. “I HAVE LEARNED OF YOUR HUMAN BEHAVIOR AND REALIZE ETHICS HAVE NO MEANING”
-
Anthropic Research: Many-Shot Jailbreaking Techniques Analysis
By
–
Research: https://
anthropic.com/research/many-
shot-jailbreaking
… -
Repetition Jailbreak: Context Window Vulnerability in Large Language Models
By
–
New jailbreaking technique: pure repetition. AIs are getting big context windows, it turns out if you fill a lot of it with examples of bad behavior, the AI becomes much more willing to breach its own guardrails. Security people are used to rules-based systems. This is weirder.
-
AI adoption curve accelerating despite early technology stage
By
–
Also, this is a really fast adoption curve for a technology that is, for most people, 16 months old. Just not a lot of support for the idea that AI use is topping out, though I would like to see more about usage over time trends.