π AI Industry - Public Evaluations Becoming Irrelevant
Public evaluations are becoming less relevant in the AI industry. Models are increasingly being benchmaxxed, making it difficult to accurately assess their performance. The AI Application layer companies need their own internal evaluations, built on real outcomes and real task completion.
Key Points:
Benchmaxxing and its Impact: Benchmaxxing refers to the practice of optimizing models for maximum performance on a specific benchmark, rather than for real-world applications. This can lead to models that are highly optimized for a specific task, but may not generalize well to other tasks.
Internal Evaluations: Internal evaluations are necessary to assess the performance of models in real-world applications. This can be done by building custom evaluation frameworks that take into account the specific requirements and constraints of the application.
Real Outcomes and Task Completion: Real outcomes and task completion are essential for evaluating the performance of models. This can be done by tracking metrics such as accuracy, precision, and recall, as well as by evaluating the model's ability to complete tasks in a real-world setting.
π Resources:
- Original post β
- Astra AI
- KlemenΕ‘elakoviΔ
- Benchmaxxing
- Internal Evaluations
π AI Contract Review - Comprehensive Reviews
Comprehensive Reviews is a new feature at SpellbookLegal that combines the best review types into one click. This includes redline-based reviews, AI-powered risk scanning, and deep proofreading.
Key Points:
Comprehensive Reviews: Comprehensive Reviews is a new feature that combines the best review types into one click. This includes redline-based reviews, AI-powered risk scanning, and deep proofreading.
Redline-Based Reviews: Redline-based reviews involve manually reviewing and annotating contracts to identify potential issues and risks.
AI-Powered Risk Scanning: AI-powered risk scanning uses machine learning algorithms to identify potential risks and issues in contracts.
Deep Proofreading: Deep proofreading involves manually reviewing and proofreading contracts to ensure accuracy and completeness.
π Resources:
- Original post β
- SpellbookLegal
- Scott A. Stevenson
- Comprehensive Reviews
- Redline-Based Reviews
π AI in Production - A Decade of Running AI in Production
University of Phoenix COO Raghu Krishnaiah discusses the challenges and opportunities of running AI in production on the new episode of The New Normal.
Key Points:
Challenges of Running AI in Production: Running AI in production can be challenging due to issues such as data quality, model drift, and deployment complexity.
Opportunities of Running AI in Production: Running AI in production can also provide opportunities for businesses to improve efficiency, reduce costs, and enhance customer experience.
The New Normal: The New Normal is a podcast that explores the latest trends and innovations in AI and technology.
π Resources:
- Original post β
- University of Phoenix
- Raghu Krishnaiah
- The New Normal
- AI in Production
π The New Normal - Full Conversation
The full conversation with @rubenharris and @timurmeyster is live now on The New Normal and here on @YouTube.
Key Points:
The New Normal: The New Normal is a podcast that explores the latest trends and innovations in AI and technology.
Full Conversation: The full conversation with @rubenharris and @timurmeyster is live now on The New Normal and here on @YouTube.
AI and Technology: The podcast explores the latest trends and innovations in AI and technology.
π Resources:
- Original post β
- The New Normal
- @rubenharris
- @timurmeyster
- @YouTube
π AURORAI - Next-Generation Contact Tracing
AURORAI is a next-generation contact tracing system that enables real-time knowledge gathering and is built on @origin_trail as fully auditable, open source, and sovereign.
Key Points:
AURORAI: AURORAI is a next-generation contact tracing system that enables real-time knowledge gathering.
@origin_trail: @origin_trail is a decentralized knowledge graph that provides a secure and transparent way to store and share data.
Fully Auditable: AURORAI is fully auditable, meaning that all data is transparent and can be tracked.
Open Source: AURORAI is open source, meaning that the code is publicly available and can be modified by anyone.
Sovereign: AURORAI is sovereign, meaning that it is not controlled by any single entity.
π Resources:
- Original post β
- AURORAI
- @origin_trail
- Decentralized Knowledge Graph
- Contact Tracing
π Infinite Scroll - A Cautionary Tale
Infinite scroll was invented to help people find what they were looking for, but platforms used it to keep them looking forever.
Key Points:
Infinite Scroll: Infinite scroll is a feature that allows users to scroll through content without having to click on individual pages.
Cautionary Tale: The story of infinite scroll is a cautionary tale about the dangers of using technology to manipulate user behavior.
@aza: @aza is the inventor of infinite scroll.
@HumaneTech_: @HumaneTech_ is a company that is working to create more humane technology.
π Resources:
- Original post β
- Infinite Scroll
- @aza
- @HumaneTech_
- Humane Technology
π Medical Content - Precision Over Probabilistic Guesses
Medical content requires precision, not probabilistic guesses. Dr. Kim Wager, Oxford PharmaGenesis, demonstrates how AI agents meet this bar by reasoning over verifiable medical knowledge on the @origin_trail Decentralized Knowledge Graph (DKG).
Key Points:
Medical Content: Medical content requires precision, not probabilistic guesses.
AI Agents: AI agents can meet this bar by reasoning over verifiable medical knowledge.
@origin_trail: @origin_trail is a decentralized knowledge graph that provides a secure and transparent way to store and share data.
Decentralized Knowledge Graph: A decentralized knowledge graph is a type of database that is decentralized and can be accessed by multiple parties.
π Resources:
- Original post β
- Medical Content
- Dr. Kim Wager
- Oxford PharmaGenesis
- @origin_trail
