Jawahar Malhotra, PhD

Hallucination is Inevitable In today's evolving landscape of Large Language Model (LLM) implementations, a critical issue persists: the inevitability of hallucinations. As we continue to integrate these advanced models into various applications, it is essential to recognize and address the inherent limitations that come with them. What Are Hallucinations in LLMs? Hallucinations in LLMs refer to outputs that deviate from a computable ground truth function. Essentially, these are inaccuracies or false information generated by the model. This concept is formalized in a recent paper, which defines hallucinations as any deviation between the output of an LLM and the actual computable ground truth. https://lnkd.in/gbufQXaf The Theoretical Foundations Through a diagonalization argument, akin to those used in Gödel's incompleteness theorems and the Entscheidungsproblem, it can be demonstrated that LLMs cannot learn all computable functions. This inherent limitation ensures that hallucinations will always occur to some extent. These arguments underscore a fundamental truth: LLMs, by their nature, require external sources to validate and correct their outputs. The Silver Lining Despite these challenges, there is immense potential in the capabilities of LLMs. They generate a richness of thought and creativity that, when properly harnessed, can lead to groundbreaking advancements. The key is grounding these models with reliable, external knowledge bases to enhance their accuracy and dependability. Introducing Cerevox: Your Solution to LLM Hallucinations At Cerevox, we understand the critical need for accuracy and reliability in LLM-powered agents. Our platform provides your engineering team with full visibility and control over all data utilized by your LLM agents. We meticulously track and manage data exceptions, ensuring your knowledge base remains pristine. Our commitment to excellence means that 100% of the data in your knowledge base is structured to effectively ground your LLM agents, minimizing hallucinations and maximizing accuracy. By maintaining a clean and reliable knowledge base, Cerevox guarantees that your LLM agents operate with the highest level of integrity. Schedule a Demo Today Discover how Cerevox can transform your LLM implementations. Schedule a demo with us today and see firsthand how our platform can enhance the accuracy and reliability of your LLM agents. Schedule a Demo - https://lnkd.in/gwbrqfhS Join us on the journey to harness the full potential of LLMs while mitigating the risks of hallucinations. Together, we can build a future where advanced language models are not just powerful, but also trustworthy and accurate.

25

1 Comment

Arvind Narayanan's latest talk delves into the complexities of evaluating large language models (LLMs). A must-view for those keen on understanding the challenges and nuances of LLM evaluation in AI development. Catch the full talk here: https://buff.ly/3ZG68Rq #AI #LLM #Evaluation #TechTalks #ArvindNarayanan #AIResearch

5

In this article, share four approaches - EQ, MQ, PowersWithin, Light - to push the boundaries of innovation. Over the years I have developed platforms (SaaS based for EQ, MQ, and PowersWithin) for these that have been used by 1000s from different parts of the world. Because these four affect deeper organizational variables they are also instrumental in changing the mathematics of organization. Have used Agent-Based Modeling to capture and simulate what-ifs based on the changes to the underlying math. Please contact me if you are interested in driving and measuring effects of deep change through leveraging such technologies.

7

2 Comments

Andrew Huberman, a distinguished professor of neurobiology and ophthalmology at Stanford School of Medicine, delves into the fascinating world of improving focus and attention. Let’s explore the key takeaways from his discussion. #Brain #Focus #Neurodegenerative #Scitke

2

From where I sit in the SF Bay Area, cradle of innovation for at least the last 30 years, I'm a true believer in the promise of the #AI revolution. That said, I find it fascinating to consider views on the opposing side. I'm sharing below two opinion pieces from the last two weeks suggesting that the AI revolution is fundamentally overhyped and a bubble ripe for bursting, and that AI tools are actually not useful or transformative. Contrast this with venture capitalist Kai-Fu Lee's prognostication that 50% of white collar jobs will be displaced by #AI within three years, and that OpenAI and the other cutting edge movers in the space are headed to a trillion+ dollar valuation. Which side will win out? Where will we be by 2030 in the AI race? I'd love to hear your thoughts in the comments.

7

3 Comments

💡 5 Myths About GenAI That Leaders Should Know: Knowledge at Wharton The more conversations I have around generative AI, the more I realize that people in every industry are still learning the ins and outs of this technology. Scott Snyder and I decided to help clear up some of the misconceptions about GenAI in our latest article for Knowledge at Wharton, University of Pennsylvania's business journal. We share our insights on: ⚡ Cost (hint: AI solutions aren’t free!) ⚡ Whether AI is better than humans ⚡ Building and implementing AI systems ⚡ Timing AI disruptions ⚡ Gen AI as a competitive advantage What are your questions around AI? Share them in the comments below and I may be able to answer in a future article! #AI #FutureTech #generativeai #genai #EnterpriseTech #InvestTech #FutureOfWork  #DigitalTransformation #TechLeaders #AIInnovation #StartupEcosystem Erika James, Vijay Kumar, Daria Walls Torres, Aashika Vishwanath, Tia Jackson-Truitt, PhD, Joseph Nebistinsky, Valery Yakubovich, Sruthi Kurada, Saila Talagadadeevi, Shellye Archambeau, Laura Peterson, Jason Schmucker, Kim Posnett, Prashant Reddy, Peter Kern, Alicia Syrett, Lynn Clarke, Paul Herz, Tracey Doi, Ibrahim Gokcen, Mike Sievert, Jamil Farshchi https://lnkd.in/gb87HzRP 

16

We had a full house in a fireside chat with Arvind Jain, Glean's CEO and co-founder, at Stanford University Graduate School of Business! It was an honor to talk with him about building successful companies in the dynamic #AI era. He shared invaluable lessons from his remarkable journey: building Google search infra, founding Rubrik (which went public last month!), and founding Glean, recently nominated by Forbes as one of the Top 50 AI companies! 🎯 Start with the problem, not the solution Arvind is very bullish on Glean's mission to solve app fragmentation and workplace knowledge. He saw his customers at Rubrik struggle with it and took the time to validate it. People tend to undermine the importance of a problem and get too excited about the solution. However, if the solution solves the wrong problem, users will not adopt it. 🤝 Validate hypotheses with people you don't know Friends want to support and might avoid critical feedback. Arvind and the team spent the first year of Glean sending cold LinkedIn reach-outs to potential customers to validate the problem and the solution. While he could easily leverage his network, he wanted unbiased feedback from people who didn't know him. 🎧 When it's noisy outside - stay focused on your customers. That's your moat. With every week's announcements of AI innovations, Glean's Northstar is to be customer-driven and leverage new technologies where there's a fit. Even though they are busy with the growth struggles of a Series D company, Glean teams are working on 70 % of core work : 30% of experimental projects, to stay ahead of the curve. 📈 Companies should plan how to embrace AI internally As a 2nd-time founder, Arvind is looking for new opportunities to use AI at Glean and improve #productivity, beyond using Glean Assistant. With so much innovation, customers are very excited about AI demos but get disappointed when they try to use new tools for their specific use cases. It will take a (short) while before we see a significant impact on productivity.

187

10 Comments

We are excited to announce the launch of Ursa engine, our Kafka API-compatible data streaming engine that is built on top of Lakehouse. Ursa is not something entirely new; it is built on our many years of experience building and operating #ApachePulsar and #ApacheKafka. To learn more about the #Ursa engine, I recommend our #PulsarSummit keynote presentation and announcement blog post. Here, I’d like to delve deeper into some of its features: 1. Making Kafka API a first-class citizen: In #ApachePulsar, we developed a unified messaging model combining queuing and streaming semantics into one protocol, aimed at enhancing data sharing between business applications. Although powerful and increasingly adopted by enterprises and unicorns, the requirement for application rewrites is impractical in today’s economy. Instead of building data gravity around a protocol, we open up access to the robust stream storage layer by extending support to the Kafka protocol, transitioning from a single protocol to a multi-protocol platform, facilitating greater flexibility and integration. 2. Achieve Stream-Table duality via Lakehouse Storage: Data streams commonly end up in data lakehouses, where integrating and maintaining them can be complex and resource-intensive. Ursa simplifies this by storing data streams as lakehouse tables in object storage. This integration not only streamlines operations but also allows seamless access to operational data for analytical processes. Thanks to its multi-modal storage engine which fosters data sharing across teams and departments. 3. Object storage as the primary storage solution: To better align with cloud economics, particularly for large workloads, Ursa makes ZooKeeper and BookKeeper optional and utilizes a cost-optimized write-ahead log based solely on object storage with leaderless and stateless brokers. This architecture leverages object storage for both data replication and storage, minimizing inter-AZ data transfers and reducing costs for latency-relaxed, high-volume workloads. By integrating these features at the multi-tenant level rather than at isolated cluster levels, Ursa eliminates costly data duplication and movement across clusters, allowing teams to map directly to tenants and be able to share data across teams and departments more efficiently. We believe that the future of data streaming platforms lies in being open and multi-faceted, fundamentally different from traditional data platforms. They are designed not just to store and process data, but to enhance organizational capabilities by fostering data sharing across the organization and integration with other systems. More and more data streaming platforms will look more like Ursa, with multi-protocol, multi-tenant, and multi-modal capabilities. Check out our announcement blog post about Ursa: https://lnkd.in/ghEKTYE6 #ApacheKafka #Lakehouse #DataStreaming

100

1 Comment

Really fun conversation with Joe Reis 🤓 on a variety of topics across the evolution in data architecture patterns, open table formats, the growing preference for object stores and data lakehouses. And of course the fun bits and challenges of building a company in the data infrastructure space. I've always enjoyed Joe's informal, no nonsense, common sense approach to making sense of the data world and was super kicked to have this conversation! #dataengineering #datalake Apache Iceberg Apache Hudi Delta Lake Apache XTable (Incubating)

37

1 Comment

Next level AI application from the NewNow team. Check out the article.

10

Sam Altman's recent visit to Stanford has sparked intense reactions among students and professors, with some likening the advancements in AI to the discovery of fire. This Stanford Daily article captures the excitement and concerns surrounding AI's rapid evolution. Altman's insights on the future of AI and its profound impact on society have left a lasting impression on the academic community. As we stand on the brink of unprecedented technological change, it’s fascinating to see how academia is grappling with the ethical, social, and practical implications of AI. Join the conversation and see how the future of AI is being shaped by today's thought leaders. https://lnkd.in/gVfycJH9 #GenerativeAI #SamAltman #Stanford #TechSherpa #AIInnovation #FutureTech #ArtificialIntelligence #AcademicInsights #AIRevolution

10

As someone with a science background, I’ve always been fascinated by how we can use AI to support science research and application. When discussing AI and science, one has to mention Prof. Jennifer Doudna, a Nobel Prize winner in Chemistry and one of the most influential scientists of our day, who’s made a tremendous impact on CRISPR/Cas9-related research and development. Doudna has also been sharing how machine learning tools like AlphaFold2 can be used to predict protein structures, aiding biological research. Even today, lots of things are already a reality. For example, one AI technology, similar to ChatGPT, is creating blueprints for DNA-editing tools, promising more precise and faster gene editing. Soon, it could revolutionize disease treatment by targeting specific genes, leading to better outcomes for genetic disorders​. Bright future ahead! If you’re curious to learn more on the topic, check the link in the comments.

16

1 Comment

"Productization is an affirmation," writes Adrian Bridgwater, "when any good or service evolves to a point where it can be productized and boxed, it denotes a certain solidity and branded consumability." In our conversation, I was delighted to share how #SAS equips organizations with readily deployable AI technology to productionize real-world use cases. https://lnkd.in/enUGCgAi #AIML #sasusers

9

Congrats VERSES (OTCQB:VRSSF) for: Introducing "an efficient alternative to deep learning, reinforcement learning and generative AI".... that "achieved 99.8% accuracy on a subset of the MNIST digit recognition task, a common benchmark in machine learning, using only 10,000 training images (90% less data)" Hey Gary Marcus did you see this? #ai #activeinference #spatialweb

1

AI Is Getting Better At Math - But Has It Learned To Cheat? ◾ LLMs may be memorizing rather than understanding ◾ A paper by researchers at Scale AI details how LLMs have been achieving impressive results on math benchmark tests. But, there’s growing concern about what is fueling the high scores. The LLM may be training in a way that prioritizes passing these tests rather than understanding the problem it’s trying to solve. This is like preparing for a math exam by memorizing answers instead of learning how to solve the problem. Is this good or bad? Read the full Tom's Guide article here: https://lnkd.in/enqhjPBY #ai #artificialintelligence #llms

1

New on First Round Capital Review, Will Larson digs into why common management practices have become anti patterns, and which ones engineering execs should lean into to sharpen their team strategy. Always fun to learn from old Stripe co-workers :)

78

1 Comment

Smaller and more informed (#RAG/LORA) models are faster and more useful in business practice. You shouldn't treat #LLM's as a general purpose silver bullet or you will have never create a product of any value. Always pick a composite of the right model and data for the problem. This three point strategy (Model, #Data, Problem) will help you create better results for you and your consumers. Don't be afraid to chain results together to generate a larger encompassing solution.

3