https://
benn.substack.com/p/we-dont-need
-another-sql-chatbot
…
@_ganeshp
-
We Don’t Need Another SQL Chatbot
By
–
-
LLMs and Query Writing: Fundamental Mismatch Between Probabilistic and Deductive Approaches
By
–
..,LLMs and query writing are fundamentally mismatched. The former is probabilistic, creative, and inductive—it’s best used to generate ideas from short prompts. But analysts have to do the opposite when answering questions. They need to be precise, rigorous, and deductive..”
-
AI Users: More Storytellers Than Builders in Tech
By
–
@magicaltome has 10M users using AI to build PowerPoints, @github Copilot has 1.5M developers! More storytellers than builders, that’s where the problem is..
-
Building a Better World with AI: Collaboration and Innovation
By
–
So, let’s build, iterate and make the world better a better place – You and #AI can change the world! And if you found this thread useful, retweet so others can see it, and follow @_ganeshp for more like this: https://
x.com/_ganesh -
AI as Tools: Choosing the Right Solution for Enterprise Problems
By
–
Lastly, it’s important to note that #AI is a bag of tools for a bag of problems, knowing what tool to use to solve what problem makes all the difference! here is a framework to think about #GenerativeAI for enterprises: h/t Alex Ratner
-
ChatGPT Performance Degradation Possibly Intentional Fine-Tuning
By
–
At least one area “Answering sensitive questions'' performance degradation may be due to purposely fine-tuning! The reality is that nobody can replicate or validate these results (you can’t unless you are at OpenAI and have access to the March version of ChatGPT)
-
Autonomize AI: Smart Vector Embeddings for Healthcare Domain Retrieval
By
–
Case in point: At Autonomize AI we use smarter ways (use vector embeddings + smaller (healthcare) domain trained models for retrieval augmentation) because at the end of the day for search/IE/business use cases, relevance and accuracy is paramount. Our customers love it!
-
ChatGPT Performance Decline Paper Sparks Debate Among Experts
By
–
This “#ChatGPT is losing it..” paper got all the views and highlights https://
lnkd.in/gSx95EUB. Seems to be mostly from the doomers, and the experts pushing their agendas (because they build LLMs). We still don't know what is happening here. -
LLMs as Tools: Building Enterprise-Grade AI Systems
By
–
LLMs are one of the many tools you have in #AI to solve real problems. These models are powerful next word predictions and are optimized for human-like conversational experiences, but in an enterprise context, you have to actually build enterprise grade systems
-
Generative AI Data Causes Model Autophagy Disorder MAD
By
–
Generative AI generated data can make #GenerativeAI models go MAD (Model Autophagy Disorder) – super cool paper to review: https://
lnkd.in/gN8QuxW7. It underscores the importance of Data and domain expertise in the age of #AI.