Thomas H.

Sharing this ZDNET article about the new solution Blitzy On Wednesday, the Massachusetts-based company announced Blitzy Platform, an agentic system that "batch-builds up to 300,000 lines of code in a single 8-hour inference run, cutting end-to-end development timelines from months to days." The platform interprets product roadmaps, generates documentation, and identifies final steps for human developers, writing anywhere from 50% to 100% of the necessary code, depending on the project's complexity. " ___________ #AI #aiethics #automation #LLM #softwaredevelopment https://lnkd.in/gur-yFSh

3

2 Comments

One of the hardest problems in generative modeling today is around #evaluation. How do you know if the output of your diffusion model is actually good? 🤔 If this sounds interesting, come join us as we tackle these hard problems in #genAI! Jobs linked in comments 👇 At Advex AI , we evaluate synthetic samples on two dimensions: 1. Fidelity - is the overall image realistic? 2. Faithfulness - does the generated image align with input instructions? A common challenge we face is the trade-off between fidelity and faithfulness. Current evaluation metrics such as CLIP, Aesthetic score, and RLHF often fail to give reliable results. Here's why: 1. Aesthetic scores (Q-Align, ICML '24) focus on how visually appealing an image is, but for tasks like generating synthetic data to improve downstream model performance, utility matters more than aesthetics. Similar to how augmentations introduce noise to improve robustness and generalization, synthetic images can be valuable even if they don’t score high on aesthetic metrics. 2. Multimodal models such as CLIP (OpenAI) and BLIP (Salesforce) excel at high-level semantics but struggle with low-level details like texture or sharpness. Additionally, these models are also biased towards the distribution it was trained on. 3. RLHF (ImageReward, NeurIPS '23) shows promise, but collecting large volumes of preference data is tough—especially in skilled domains like medical imaging where even the expert judgements are subjective. These limitations hinder full automation of genAI evaluation, but it doesn't have to be this way. At Advex, we’re pioneering new solutions to generative model evaluation! #SolveDataSolveAI

72

8 Comments

Interested in turning transcripts into Knowledge Graphs that you can ask complex questions across multiple Congressmen, political issues, and Committees? In this WhyHow.AI case study, we show how Congressional Committee Transcripts are turned into Temporal Knowledge Graphs that you can rely on for the purposes of RAG and analytics. We show what real Question & Answers are against this system, and what type of business outcomes you can achieve with this system. As far as we are aware, the combination of steps here is a novel Knowledge Graph implementation, and incorporates the use of unstructured.io and Senzing for data parsing. Here, we showcase all the inputs into the Process, including a step by step of the data transformation process, breakdown of the 20 dev hours involved to build the entire system end-to-end, schemas used, Questions & Responses, and a link to the Graph created. Chris Rec Thomas Smoker https://lnkd.in/eGavCPch

63

6 Comments

Read more about Curvenote and ROR 👇 “This represents the ideal of a scholarly ecosystem that is rich with APIs and structured data, and our authoring tools need to be able to show the value of that as you’re working. You see a paper, you copy the DOI, and then you don’t have to think about the references, you don’t have to think about APA formatting. You see a ROR, you copy it into your document, and then it turns into the name of the organization with all of the information directly available in a hover reference."

14

Anyone building #AI - whether you are selling it or leveraging it internally - should heed these wise words from Thomas W. Dinsmore in his latest substack post: Every model has a business stakeholder; most stakeholders aren’t data scientists. Opaque models are dead on arrival. 👏 #climatetech #climateAI

5

1 Comment

Nice shoutout from Dotan Horovits 🇮🇱🎗and Charity Majors for #OpenTelemetry and the evolution of client-side telemetry (particularly the open source of Embrace) in 2024. Time for data to be portable, available, and not locked in. "F that sh$t. The only thing that deserves swearing is vendor lock in," screamed Charity Majors. Data should be available to anyone who can build on it, run analysis, do AI, etc. Stabilize, adopt, and run!

13

1 Comment

[New Analysis][The Future of Prompting Mechanisms] A Definitive Guide for Fully Agentic Applications Link Below ⬇️ Like, repost, and share it with your network 😆

3

2 Comments

Today's article on MCMC - read to learn why there are better and more efficient techniques than chasing LLMs and other #AI buzz. https://lnkd.in/eaw6Paan

8

This is more of a "talking to myself in public" thing than any kind of fully fleshed out hypothesis, so take this with the appropriate grain(s) of salt. But if I were interested in AI and were looking for an angle to explore that might yield something interesting, I'd start working on learning "Multiscale Mathematics" (also known as "Multiscale Modeling"). This is rooted in my personal belief that there has to be a "multi-scale" aspect to intelligence because there is such an aspect inherent in our underlying physical reality.

2

Anthropic has launched Claude 3.5 Sonnet, the inaugural release in their upcoming Claude 3.5 model family. This new AI model sets unprecedented industry benchmarks for intelligence, outperforming competitor models and even Anthropic's own Claude 3 Opus across a wide range of evaluations, including graduate-level reasoning, undergraduate-level knowledge, and coding proficiency. Claude 3.5 Sonnet demonstrates significant improvements in understanding nuance, humor, and complex instructions, while excelling at producing high-quality content with a natural, relatable tone. ➜ Check the news at: https://lnkd.in/dbKMKPm7 ➜ Vendorvise's official Instagram: https://lnkd.in/ebK2KQ2C ➜ Vendorvise's Linktree: https://lnkd.in/dQdbMdth 🤖 If you enjoy reading my posts about AI and tech news, hit that like button and share with your friends to support my content! #ai #artificialintelligence #ainews #aipowered #news #blog #claude3 #sonnet #anthropic #generativeai #llms #tokens #aitools #aitech #tech #technology #hardware #content #socialmedia #future #new #launch #business #techtrends #innovation #success #markets #people #industry

2

Hot take: Models and Infrastructure are rapidly becoming commodities. The true moats are UX, Distribution, and Data. I foresee the application layer as the nexus of innovation and impact. The application layer is where the real value will be created. While foundational, models and infrastructure are now standard fare. They are critical tools that can be used to build applications. 💡 That’s why prioritizing distribution and ecosystem creation is key. True differentiation lies in capturing and leading markets through dynamic strategies. #TechInnovation #FutureOfTech #UserExperience #DistributionStrategy #DataDriven #MarketLeadership #EcosystemBuilding #BusinessStrategy #artificialintelligence #aimodels #userexperience

24

Dr. Karl-Michael Popp it was such a pleasure to talk with the great attendees at ewesco.org today. Key points from the talk: 1.GenAI code tools produce significant productivity gains, and >1000% Return on Investment from purchasing the tools 2. The risks from GenAI code are manageable via GenAI Code Transparency-- knowing what's in your codebase... 3. ... and by developers blending the code, not just accepting the "pure" results 4. Legal teams at medium and large companies need to be thinking, now, about the future Intellectual Property implications to make sure their IP stays safe as GenAI code usage increases.

18

2 Comments

🚀🔧 Sorella Labs, a startup focused on developing a MEV (Miner Extractable Value) risk-hedging tool, has raised $7.5 million in seed funding. The funding round was led by Paradigm, with participation from Uniswap Ventures, Bankless Ventures, and Robot Ventures. Sorella Labs is working on a solution to address the issues arising from the MEV strategy, where miners, block builders, and validators can insert or reorder transactions within a block to maximize rewards. 🌐🔗 #BitcoinWorld #BlockchainInnovation

1

So - are LLMs Universal? This guy looked at a lot of proofs that LLMs are universal and concluded that they are all wrong or misleading. I didn't check all the proofs and I didn't check all the errors or misunderstandings that he mentions. But I would say that it is not confirmed that any realistic model of LLMs is universal. Specifically about https://lnkd.in/da4qHFgh I did not make up my mind if the model is realistic or not, but I think it makes improvement by 1) Getting the outer while loop into the LLM by emphasizing auto-regression 2) Getting the ability to skip parts of the input with the bounded context window, whether a realistic aspect of a theoretical model or not.

4

1 Comment

AI in Local Government: City of Boston's Interim Guidelines for Using Generative AI (Part 2 of Series) Overview: The City of Boston has taken a proactive step in addressing the use of AI in local government by releasing interim guidelines for using generative AI. These guidelines aim to encourage responsible experimentation while managing potential risks. Target Audience: All City of Boston employees and agencies (except Boston Public Schools) Key Features of the Guidelines: • Principles for AI use (e.g., Empowerment, Inclusion, Transparency) • Specific guidelines for content generation and review • Sample use cases and examples using ChatGPT and Google's Bard • Do's and Don'ts for various AI applications • Resources for further learning Standout Elements: • Encourages responsible experimentation • Addresses privacy and security concerns • Provides actionable examples for common municipal tasks • Emphasizes transparency and accountability in AI use Key contacts: • Santiago G., Chief Information Officer, City of Boston • Department of Innovation and Technology: [email protected] Personal comment: Boston's AI guidelines are one of the first publicly released by a U.S. city. What impresses me is the balance between innovation and responsibility. It's not just policy; it's a practical toolkit with real-world examples using ChatGPT and Bard. For cities wanting to experiment with generative AI but needing guardrails while developing comprehensive policies, this is an invaluable resource. It's refreshing to see a city embrace new tech thoughtfully, with an eye on inclusivity and transparency. Boston's approach offers a solid starting point for any municipality navigating the AI landscape. Link to Boston's AI Guidance: https://lnkd.in/gTDhCh8w #AIinGovernment #SmartCities #ResponsibleAI #BostonInnovation #AIPolicy

10