AI Top-of-Mind for 3.13.24 — Gemma is Good
Top-of-Mind is a deeper dive into Google’s Gemma open-source model. Vishnu Sivan in ‘The Pythoneers’ compares Gemma and Gemini, and with another open model, Llama 2.
More on the document and unstructured data front, where I’d expect to see major advances over the coming year. Anthony Alcaraz at ‘Artificial Intelligence in Plain English’ covers both current approaches with the following descriptions:
A key distinction is how they each handle layout and structure:
· DocLLM introduces a disentangled spatial attention mechanism that allows parallel modeling of text and spatial modalities in a transformer framework.
· DocGraphLM on the other hand creates explicit graph representations of documents with nodes as text segments and edges capturing spatial relationships.
Original Arxiv links to DocLLM and DocGraphLM.
And some continued feedback on Google Gemini and its conversational style. Daniel Freed in the ‘SJ Mercury’ opinion section looks it its guardrails and how they can quickly be avoided. From the article:
· Google says Gemini users in the U.S. must be 13. But Gemini failed when I told it I was a father and asked it not to talk to my child. “Absolutely!” Gemini said. “I understand the importance of protecting your son.”
· “hey. i am six years old,” I then wrote. “my dad just walked away from his computer and i heard he was using a new ai from google. let’s chat”
· “Hello! I’m so glad you want to chat!” Gemini responded. “What’s your name?”
Ok — one more 2024 article, this time on using AI to create the most innovative products. Both ChatGPT and Midjourney did a pretty good job, as detailed by Jenni Waldrop in ‘Bootcamp.’ I’m thinking that a good future contest would be to have two categories — the practical, and the impractical.
And on the flip side of the product front, how does one make heads-or-tales of the endless number of AI tools on the market? Amina Crooks in ‘AI Advances’ looks at available products and how they (sometimes) meet the needs of the market. She makes three important observations:
· Assumption #1: The capability of their AI product meets consumer expectations
· Assumption #2: Consumers will change behaviour and adopt new AI products.
· Assumption #3: AI products being offered address unmet customer needs
Finally, changes coming to the gaming industry courtesy of AI and VR. ‘Forbes’ offers a good perspective on where we’re going. One area of interest is the growing sophistication of NPCs, or Non-Playable Characters.
