Kate Rosenshine

AI companions are no longer just a sci-fi fantasy but a growing reality with potential risks we must address. As we increasingly welcome AI into our lives as friends, lovers, and mentors, we must ask: are we trading the messy beauty of human connection for the seductive ease of AI companionship? This thought-provoking article delves into the potential dangers of AI addiction and proposes innovative regulatory approaches to safeguard our well-being. It's a must-read for anyone interested in the future of human-AI interaction. #AI #addiction #regulation #ethics #humanconnection #EmergingTech #GenerativeAI https://lnkd.in/gbFXpJYC

1

As data and AI have exploded, the ability to process and refine data rapidly and efficiently, as well as keeping a handle on governance and quality, has put immense pressure on enterprise data teams. Orchestra was founded in 2023 by the inimitable team of Hugo Lu and Will Davies to solve exactly that problem. It offers a unified control plane for data operations, enabling users to build data pipelines and data products, automate pipeline execution, and gather metadata in a governed way – without the need to maintain infrastructure. That’s why we at Moonfire are leading Orchestra's £1m pre-seed, alongside Sequoia Capital’s Scout Fund and angel investors from companies like dbt Labs, Snowflake, and Snowplow. You can find out more about Orchestra via Pulse, in the comments below. ⬇️

81

3 Comments

Smaller is better The need for powerful #LLMs has fueled their exponential growth. However, large LLMs come with a cost - they require significant computational resources and energy, leading to increased response latency. The largest models today have over 100 billion parameters and can take over 200 gigabytes to load. Industry applications and #AI giants are now shifting their focus towards smaller and faster models, prioritizing efficiency over maximum performance. Learn more:

13

Over the last couple of years Large Language Models (LLMs) have emerged as a disruptive technology with potential to transform healthcare, however the accuracy and reliability of information provided by LLMs can be a matter of life or death, as it could potentially affect healthcare decisions, diagnosis, and treatment plans. In order to address this issue The Open Medical LLM leaderboard provides a comprehensive assessment of each model's medical knowledge and question answering capabilities https://lnkd.in/eusBVbhB

40

Individuals are increasingly turning to #AI in both their personal and professional lives, yet many businesses struggle to fully harness its potential. My colleague, Colette Stallbaumer, spoke with TechRadar about the need for companies to develop a business strategy for AI, not just an AI strategy. By integrating AI into business strategies, organizations are going to begin seeing tangible, bottom-line results and cost savings. Colette offers some valuable tips to help individuals and organizations alike find value in AI solutions like Microsoft #Copilot, enabling them to tap into time and productivity gains and boost creativity. Check out her interview here: https://lnkd.in/g5Ukv5-T

185

6 Comments

Gartner’s #AI #Leaders team: “Navigating #GenAI #Production and #Evaluation Challenges” Pleased to join our recent roundtable, with select peers in the AI Leaders group, on the above very engaging topic, expressing my personal views. We explored: Understanding GenAI production strategies * How do you define and prioritize best practices when implementing GenAI production strategies? * What considerations are essential when selecting workflows for managing GenAI models? * How do you balance the need for innovation and experimentation with the necessity of maintaining stability and reliability in GenAI production? GenAI model evaluation * Beyond traditional metrics, what additional factors do you consider when evaluating the performance of GenAI models? * How do you monitor and evaluate the performance of GenAI models in production for drift or degradation? * How do you approach the validation and verification of data sources used to train GenAI models? Integration Hurdles * What are the primary challenges you've encountered when integrating GenAI production into your existing development pipelines? How are you overcoming them?
 Can you share any "aha" moments or unexpected discoveries that have occurred during integration efforts, and how have they influenced your approach going forward? * Are there any regulatory or compliance considerations that have posed challenges during integration, and how do you address them? * What role does change management play in successfully integrating GenAI production into existing workflows, and how do you ensure smooth adoption by end users? As always, such conversations with senior leaders practicing AI and Generative AI are thought provoking and, in my view, enable us to go beyond the typical expectation of what is possible. Thank you Gartner, for hosting this event, and for your kind invitation. #GenerativeAI #Strategy #LLMs #Innovation #transformation #artificialintelligence #AI #GenAI #ChatGPT #disruptivetechnologies #cxos #ceos #cfos #ctos #caios #cios #cdos #excellence #GenAINY

67

8 Comments

As companies from diverse sectors are starting to leverage large language models (#LLMs) and generative AI (#GenAI), LLMs’ lack of domain-specific context can create a cascade of challenges. In this article, Aerospike Chief Scientist Naren Narendran delves into how vector databases and embeddings are a powerful antidote to these challenges, and can greatly enhance search accuracy. https://lnkd.in/eAWNDZVc

7

Ben Lorica 罗瑞卡 and I continue to write and speak about similar #AI topics. After I wrote https://bit.ly/46M9gi6 we decided to have another on-air conversation about #generativeAI in the #enterprise. I always look forward to these conversations (public and private) with Ben. His in-depth knowledge of the subject and his penetrating questions make them a learning experience.

6

2 Comments

Unlocking the Power of LLMs: a framework for adoption Last week at Ai4 - Artificial Intelligence Conferences I participated in the Making Use of LLMs panel, moderated by AI legend Manoj Saxena, with industry leaders Christopher Stephens, Dan Mackey  and Raj Aikat. We discussed various strategies effectively leveraging LLMs in practical applications. My contribution focused on the progressive adoption model Scotiabank is using to walk/crawl/run with LLMs. Each successive stage in this model is more complex to implement than the previous one, but also yields greater benefits. The stages are: 1. Public Chatbots: While limited in their capabilities and requiring controlled use, platforms like Microsoft Copilot and ChatGPT can be valuable tools for numerous applications. 2. General Productivity Tools: Tools like Copilot 365 (and others) access your documents, emails, and chats, transforming how we work. They can draft emails, generate code, and summarize complex information, significantly boosting productivity. 3. Custom Tools for Specific Teams: this is where it gets exciting! Scotiabank is developing LLMs tailored to the needs of specific employee groups, saving time and boosting efficiency. 4. Back-Office Automation: LLMs are also revolutionizing back-office processes. For example, imagine a chatbot summarizing customer interactions before transferring the chat to a live advisor – this streamlines workflows and improves customer experience. At Scotiabank, we believe LLMs have the potential to transform every aspect of our operations. By strategically implementing these powerful tools, we're driving innovation, enhancing productivity, and creating a more efficient and customer-centric experience. Photo credit: César Francisco Landa Paredes

251

12 Comments

Harvey, a trailblazing AI-powered platform for professional services, has successfully raised $100 million in a Series C funding round. This significant investment, led by GV (Google Ventures) with participation from OpenAI, Kleiner Perkins, Sequoia Capital, Elad Gil, and SV Angel, values the company at an impressive $1.5 billion. Congratulations to co-founders Winston Weinberg and Gabe Pereyra for continuing your successful push forward. Harvey has quickly become indispensable to tens of thousands of lawyers, professional service providers, enterprises, and governments worldwide. Over the past two years, the company has seen its ARR triple and its workforce double, highlighting the immense value and versatility of its platform. This latest funding will enable Harvey to scale its operations, enhance its AI capabilities, and expand its global footprint. The Series C round will be pivotal in advancing Harvey's mission to build the most trusted AI platform for professional services. The new capital will be invested in #engineering, #data, and domain expertise to develop AI-native systems that address complex knowledge work. Harvey plans to grow its team, refine its domain-specific models, and deepen partnerships with #cloud and model providers to integrate and enhance its AI offerings. This funding underscores Harvey's role as a pioneer in #AItechnology, driving more effective decision-making in professional services. The continued support from clients, validation from investors, and dedication from the team inspires confidence in Harvey's future as a catalyst for progress in the industry. #AI #ProfessionalServices #StartupFunding #Innovation #TechNews #VentureCapital #Entrepreneurship #TechEcosystem #LegalTech #Startups

2