You are currently viewing AI Model Upgrades, Cost-Cutting Tips, and U.S. Election Insights

AI Model Upgrades, Cost-Cutting Tips, and U.S. Election Insights

Hello AI Friends,

At this week’s AI meetup we had a small but mighty turnout with an array of expertise using AI in their businesses from real estate, electric vehicle charging, and Meri, an ML Engineer dropping by from SF.

Some serious knowledge in the room this week.

The headline this week is the upgrade to Claude Haiku – only available at the moment through API call – but which is set to be a serious competitor to GPT 4o mini.

As usual the benchmarks paint a rosy picture but it remains to be seen if it has outpaced 4o mini.

There’s also valuable tips included this week on how you can reduce your subscriptions to various AI platforms if you’re interested in experimenting with AI images and video: a somewhat costly exercise. 

And of course since we were meeting on the eve of the U.S election, there was some news from Perplexity and their approach to engaging with what is surely going to be a fascinating night, whatever the outcome.

Enjoy and I’l see you at the next one.

Harry 

Attendees

  • Harry Verity (Host, Tech Journalist)
  • Will M (Product Marketer, joining from Australia)
  • Kesha (AI Story Platform and Developer working with Bollywood)
  • Developer from Canada (working on EV chargers)
  • Meri (ML Engineer visiting Bali from San Francisco)
  • Other participants including an Unnamed real estate professional

Key Topics Discussed

1. Claude Model Upgrades

  • Claude 3.5 Sonnet and Haiku upgrades
  • Anthropic’s claim that Haiku outperforms Claude 3 Opus on many benchmarks

Opinions:

  • Harry: Claude is now generally better than ChatGPT for most use cases
  • Will: Impressed with Claude’s improved reasoning ability and thorough responses

2. Search GPT intergrated into Chat GPT

Opinions:

  • Some skepticism about LLM technology being the best for search functionality
  • Concerns about decontextualization of information in AI-powered search

3. Grok AI (Elon Musk’s model) Demo

Opinions:

  • Mixed views on Grok’s competitiveness and Elon Musk’s claims about development speed
  • Skepticism about the promised release of Grok 3 by year-end

4. OpenAI’s O1 Full

Opinions:

  • Interest in the expanded capabilities, but skepticism about solving complex puzzles
  • Discussion on the potential internal architecture of the model

5. AI and the U.S. Election

Opinions:

  • Appreciation for the interactive features
  • Concerns about potential biases and the challenge of real-time result calling

6. Suno AI Music Generation

Opinions:

  • Excitement about the creative possibilities
  • Discussion on the lack of transparency in the underlying technology

7. Runway Gen 3 Updates

Opinions:

  • Recognition of Runway’s rapid progress in video generation capabilities
  • Discussion on the potential impact on creative industries

8. Fal AI Platform

  • Introduction as a cost-effective alternative to multiple AI subscriptions
  • Ability to buy credits to use for several different models including Runway Gen 3, FLUX.AI and FLUX LORAs as well

Opinions:

  • Positive reception as a potential solution for accessing multiple AI models
  • Questions about the long-term viability and pricing model

9. Product Marketing with AI (Presentation by Will)

  • Framework for effective AI prompting in marketing contexts
  • Demonstration of AI-powered competitive analysis

Opinions:

  • Appreciation for the structured approach to AI implementation in marketing
  • Suggestions for improving the presentation and making it more accessible to non-technical audiences

Leave a Reply