I am just back from a Busman’s holiday in Finland with a week of work and then a week exploring Helsinki and a trip over to Tallinn across the pond in Estonia with my wife and daughter. A beautiful part of the world and well worth a visit.
Some of my former colleagues / friends [they have not kicked me out of the WhatsApp group yet] were the core team that built the JPMC Research Analyst chatbot, great people build great things.
This a really good summary for anyone of what are the main trends in GenAI over the last 6 months and a view to the next, it sharp, concise and completely on point.
https://blocventures.com/generative-ai-evolution/
OpenAI going looking to drain Googles moat by launching into search with SearchGPT
With a reported 5 billion in losses for OpenAI it might be in a race against time to get past that Google moat
Google has been holding an internal prompting competition called Golden Prompts to help improve the efficiancy of its LLMs like Gemini. The Winning prompt was
“Every time I get a new Google review from a customer, add it to my Reviews Sheet, and suggest a response. Let me review the response before publishing to Google Maps.”
https://fortune.com/2024/07/25/sundar-pichai-use-cases-bots-google-competition-prompts-prizes/
CrowdStrike has released the RCA on the issue that broughs large swathes of the worlds IT systems to their knees and it turns out [*drum roll] they need to improve their QA and release process.
TLDR - Indiscriminate use of AI-generated content for training leads to "model collapse," degrading AI performance over generations, losing data distribution accuracy.
https://www.nature.com/articles/s41586-024-07566-y
They say the eyes are the windows to the soul and when it comes to differentiating between photos of humans and AI generated images of humans it appears that tools used by astronomers pointed at the eyes can tell the difference. Getting serious Blade Runner vibes.
The big players trying to play catchup with the open source community on small language models
The results of Sam Altman’s basic-income study are in and show increased spending on essentials like food, rent, and transportation, but no significant improvements in health outcomes. Participants reportedly experienced reduced stress and greater financial stability initially though this reduced over time
https://www.businessinsider.com/sam-altman-basic-income-study-results-2024-7
Microsoft will not be releasing its latest AI speech generator as its too realistic