Exploring AI: Demos, Applications, and Resources

ViewsAct Project

  • Request for demo video
  • Creator opted out due to use of real-world company designs and data

Hackathon Demos

  • Discussion on winners in SF/NYC/Berlin circuits
  • Request for more information on tracking

Dalle Short Film

  • Link shared: https://www.instagram.com/reel/CqstiaNuSjX/?igshid=MDJmNzVkMjY=
  • Another link shared for breakdown process

Networking

  • Request to meet people working in gen AI space in Bay Area
  • Introduction of Applied Scientist at Amazon building computer vision tech for video action recognition

AI/Art/Gaming Event in London

  • Link shared: https://www.theguardian.com/games/2023/apr/06/now-play-this-ai-video-game-somerset-house-london?CMP=Share_iOSApp_Other

WebGPU

  • Chrome has shipped WebGPU

Pichai and Google

  • Discussion on Pichai’s involvement in Chrome and web acceleration ideas
  • Hope for battery tech to keep up
  • Mention of Google Keyboard and Microsoft’s acquisition of SwiftKey

Agent Planning

  • Discussion on evolution of agent planning
  • Question on improvement in GPT4

Running Models on Personal Machine vs API

  • Question on benefits of running model on personal machine
  • Response on lower latency and higher throughput
  • Mention of compute cost for companies serving consumers at scale
  • Possibility of freemium models

pgVector

  • Discussion on limitations of pgVector for LLM apps
  • Request for documentation on NN capabilities
  • Mention of recall in the area of 50%
  • Suggestion to test/verify empirically
  • Question on why pgVector won’t work
  • Mention of ann-benchmarks by Erik Bern of Modal Labs

Other Vector Stores

  • Discussion on the logic behind having multiple vector stores
  • Mention of LangChain components and cloud solution for monetization
  • Mention of Pinecone being ahead in technology and supporting largest number of use cases
  • Concerns about competition from Azure, AWS, or GCP

Learning Resources

  • Links shared for Rime AI, foundational models, and database internals
  • Idea for generating embeddings and creating top 10 learnings from group every week

The description and link can be mismatched because of extraction errors.

  • https://www.instagram.com/reel/CqstiaNuSjX/?igshid=MDJmNzVkMjY= : The message mentions a shortfilm using dalle and asks if anyone has seen it. The given URL is for the first shortfilm using dalle.
  • https://www.theguardian.com/games/2023/apr/06/now-play-this-ai-video-game-somerset-house-london?CMP=Share_iOSApp_Other - For those in London: this might be worth checking out. Lots of AI/art/gaming stuff.
  • https://twitter.com/mathemagic1an/status/1644123645432958976?s=46 - The tweet discusses the possibility of freemium models with no strong QoS and asks whether it is Pichai or Larry/Sergey who will make the decision.
  • https://twitter.com/wileycwj/status/1644220882062282752?s=46&t=lkuvFQUWr1nav0QpUpFmdQ: A tweet discussing the effectiveness of pgVector and its limitations at certain scales.
  • https://twitter.com/jobergum/status/1643187540222959616 - Good discussion about pgvector in this thread. Also, a question about trying out Midjourney’s /describe feature (img2text2img).
  • And avatars come for news anchors: https://twitter.com/KhaleejMag/status/1641123893145485343 (The message in the same link as the link is related to the link.)
  • https://www.linkedin.com/posts/hsemina_generativeai-community-hackathon-activity-7050057219123400704-NIHR?utm_source=share&utm_medium=member_android - The message in the same link as the URL expresses skepticism towards benchmarking blogs by other vectorDB startups and expresses hope for the success of pgvector within the supabase ecosystem. The link is related to the context of the message.
  • https://medium.com/m/global-identity-2?redirectUrl=https://blog.startupstash.com/everything-i-wish-i-had-known-about-raising-a-seed-round-a615f8f7740b: A message asking about the experience with Fixie, mentioned in the same link.
  • https://medium.com/m/global-identity-2?redirectUrl=https://blog.startupstash.com/everything-i-wish-i-had-known-about-raising-a-seed-round-a615f8f7740b: A message asking about the experience with Fixie, mentioned in the same link.
  • The last bad product I used was https://www.steamship.com (platform to deploy langchain applications).
  • https://twitter.com/nirantk/status/1644290469915164672?s=46 - Relevant to the conversation, the user loves the quadrant.
  • https://rime.ai/ - A link shared by someone who loves learning database internals, along with a related tweet from Lily Clifford on Twitter.
  • https://rime.ai/ - This is the main website for Rime, but there is no direct context provided in the given URLs. However, the two Twitter links may be related to Rime in some way.
  • https://twitter.com/lilyjclifford/status/1643702014680133632 and https://twitter.com/younesbelkada/status/1644341068241186818?s=46&t=icC0fizZK8E3ONsDVuGFWA: The message in the link states that new foundational models for different tasks are released everyday.
← Innovative AI Applications in Smart Homes and Game Development
AI Chat Discussions and Shared Resources →