Stuart Gillen

Digital computation will not bring artificial intelligence Scientists from Stanford university describe in simple words why the current digital technology is inadequate for AI: "The place where computing went wrong, unfortunately, was the digital decision," Surya Ganguli, an associate professor of applied physics at Stanford, told scientists, academics, and other experts gathered at the HAI at Five conference today. "We decided to store information in bits which were in turn stored and flipped by shuttling many, many electrons around through complicated transistor circuits. Every fast and reliable bit flip requires, by the laws of thermodynamics, a large energy expenditure. So we expend a lot of energy in the intermediate sets of the computation. "Biology is completely different. The final answer is just good enough and all the intermediate sets are slow, noisy, and unreliable. But not so unreliable that the final answer isn't just good enough for what's required ... So I think we have to rethink the entire technology stack from electrons to algorithms in order to really go from megawatts to watts." ——- Energy expenditure is only one dimension of intelligence. It’s is important but not necessarily critical. What is becoming evident with the latest in digital technologies, specifically in GPUs, is that the current digital model is not viable for artificial general intelligence or AGI. If we were to follow the exponential diagrams pictured in one of the singularity advocates, Ray Kurzweil, we should have been able to create the intelligence of small mammals. And we are not there yet. So far what digital AI is demonstrating is that: * requires a lot of data * that data needs a lot of pre-processing * the algorithms require immense computational power of the digital kind * and this computational power requires lots of electricity, enough to start competing with human requirements Digital computation was born by the necessity of war, by evolving from the relay switch, to the vacuum tube, to the transistor, then, to the modern chips, which use electrons to define the binary state. There is a limit to what we can do with digital. And with days it’s becoming evident that artificial intelligence has hit a wall. Stanford it’s possibly one of many scientific institutions that has realized about it and we need to come up with a new computational model. #AI #artificialIntelligence #science #digitalComputation

9

AI: the bubble that burst What’s wrong with the current unscientific AI. Let’s summarize the WSJ article: * Uncertain return to investors * Unrealistic goals * Not delivering on its promises * Narrow set of practical applications * Many model variations of the same algorithm * Taking a long time, or never, to realize the business case * Spending does not guarantee a return * Pace of improvement slowing, or too many errors * Imperfect data; data being appropriated from others; * Open source options catching up with paywall models * Excessively expensive to run and maintain for meager results * Public starting to realize of its limits: self-driving cars fatal accidents; poor search results; lawsuits by owners of the data; driven by personalities not science; increasingly taking more electrical power from the grid Finally, at the end of the article, WSJ closes with: “The problem is that the current level of investment—in startups and by big companies—seems to be predicated on the idea that AI is going to get so much better, so fast, and be adopted so quickly that its impact on our lives and the economy is hard to comprehend. Mounting evidence suggests that won’t be the case.” Source: The Wall Street Journal https://lnkd.in/gg-7hJjN #artificialIntelligence #ai #machineLearning #dataScience #spe #petroleumEngineering

155

61 Comments

Current Time Series Anomaly Detection Benchmarks are Flawed and are Creating the Illusion of Progress: https://lnkd.in/gFKSqrH6

1

Here is another misuse of Artificial Intelligence If the science and physics of the state of the elements (solid, liquid, vapor, and plasma) is already settled and understood, why people insist in using machine learning to rediscover what we alll already know? To me it is a symptom that pseudo-AI companies are pushing researchers to use their software to recreate physics with only data. I have seen this happening in respected oil and gas companies. Usually this occurs when the IT departments take too much control of the data science and machine learning workflows that engineers use every day. My message is this: do not succumb or fall in the temptation of so-called AI shiny software. Know and learn you science and physics first. Keep in mind that marketing or sales people are not in their majority engineers or scientists. #artificialIntelligence #AI #spe #petroleumEngineering Massachusetts Institute of Technology

40

5 Comments

I've been checking this out at the Dianomic booth and I’m working on getting an evaluation license so I can run some data through it myself. Data quality is always an issue. I’m exploring using FogLAMP as a data path for critical environmental data. Intelligence at the edge allows me to monitor the data as it is collected and alert for issues. This plugin should simplify the monitoring and be more effective. I also have the option to buffer the data for more extended periods insuring a fail safe backup. #AVEVAWorld

13

2 Comments

"AI won’t take your job, but the person using AI will." 💡🤖 As AI becomes more integrated into our workflows, it’s clear that the real value isn’t in the tech itself—it’s in how we use it. Those who embrace AI will have a significant edge, able to do more in less time, analyze deeper, and make smarter decisions. 📊✨ In any business or domain , this isn’t about replacing people; it’s about empowering them to achieve more. To remain competitive today , use AI.📈🚀 #ArtificialIntelligence #FutureOfWork #Consulting #Innovation #Adaptability Javier Adrian Lopez Ralph Aboujaoude Diaz Pascal BORNET Ibrar Ulhaq

19

🔑 Unlock the Power of MATLAB and Simulink with Databricks! 🔑 Engineers and data enthusiasts, mark your calendars for #MATLABEXPO on Nov. 13-14! Dive into seamless data science and engineering workflows that integrate MATLAB, Simulink, and Databricks. 🌟 What You'll Discover: >> Efficient data processing and analytics << >> Advanced modeling and simulation techniques << >> Real-world applications and success stories << 💡 Why Attend? Enhance your data science and engineering capabilities by learning how to leverage the combined power of MATLAB, Simulink, and Databricks for innovative solutions. 👉 Learn More and Register: https://spr.ly/6047qp9hL #DataScience #DataEngineering #Simulink #MATLAB #Innovation #Engineering #MathWorks

4

AI Agent Systems! As we discussed with the Compound AI Systems blog in Feb, we're seeing this in practice. Moving from Monolithic to Modular design is a natural engineering progression..it means we can control quality, latency, etc for each app in Databricks! https://lnkd.in/gFacmiwR

305

1 Comment

A good post. Highlighting the "actual" areas LLMs are winning at.

4

Turns out #GenAI takes some basic common sense: (1) use the right Model for the right Use-Case (2) use Open models to leverage the Open eco-system (3) run models - where they should run - based on data availability, proximity, latency, operational efficiency, etc... On Tuesday, July 23, #Meta announced the launch of the Llama 3.1 collection of multilingual large language models (#LLMs). The instruction-tuned Llama 3.1-405B, which figures to be the largest and most powerful open-source language model available today and competitive with the best proprietary models on the market. So, if you have a use-case that benefits from this large scale OPEN model - it is now available on IBM® #watsonx.ai - where it can be deployed on the IBM cloud, in a hybrid cloud environment or on-premises. #watsonx provides a consistent common platform for use with your models, open models, and our models.... Read the announcement here: https://ibm.biz/BdKqta

18

Organizations have insufficient technical capabilities for the speed and scale of their ML model implementations. My colleague Sumit Agarwal just published a very helpful research with a capability view and sample vendors for defining a modular, flexible and scalable MLOps solution. https://lnkd.in/gnB_T9bq #MLOps #LLMOps #AI

149

17 Comments

Why I am not surprised “Further down the value chain, away from the glow of Nvidia, lurk signs of discontent. Businesses have cut back on whizzy new AI tools out of concern for hallucinations, cost and data security. “ “First, tech companies need to identify where their hype machine has gone wrong. They didn’t set expectations for AI’s capabilities too high; they framed its use as being too general purpose.” “Framing AI as a general-purpose Swiss Army knife for productivity inevitably leads to paralysis for its end users: Where do you even start with a technology that can do everything?” “AI isn’t yet a jack-of-all-trades but a master of a few. The sooner business leaders realize they can apply it to an array of niches and not for everything, everywhere, all at once, the sooner they can make the technology useful for them.” *** What prompts this analysis by Bloomberg is the meteoric rise of the stock price of NVIDIA. It is producing great concern because it resembles the internet bubble we saw 20 year ago. There is no basis for the 3 trillion USD value of the GPU company. How do we start correcting course: 1. Stop calling data science and machine learning AI. They are not interchangeable terms 2. Treat AI as a science not a mercantilistic goal 3. Don’t buy “AI” solutions for the sake of being trendy and smart. You will get the opposite effect 4. Start slow. Build your foundations with statistics, data science and data quality processes. Examine the need for machine learning algorithms 5. Inform yourself about what AI is. Put your hands on an old good friend: books. 📚 Be extremely selective with your internet choices of reading, mindful of the internet press makes money by your clicks and your data 6. Set realistic goals and establish a real, concrete result if you get hooked with “AI”. If a vendor is too insistent, try a demo or a pilot of the “AI solution” for six months, at no cost 7. Ask the vendor for real world results. Data is not perfect. You may want to try the “AI” product on your data first 8. Move your focus to the data. See if your project or challenge qualifies for machine learning. If your data is imperfect, solve the issue with the measurement processes upstream. No algorithm or “AI” does miracles. It’s all about the data 9. Be specially cautious if the vendor claims to replace your “old” and “slow” physics based models, or sound engineering workflows, with data-driven only, machine learning, or solve-it-all super fast “AI” 10. Digital Transformation is not about switching old toys for new shiny toys; it is about improving efficiencies, solving business problems, make the intricate reproducible. There is no point spending $100 million to obtain a benefit of $1 million 11. Use common sense. Ask your engineers and end users on the “AI” product. Involve them, use the bottom-up approach, and not the power of your title #AI #artificialIntelligence #spe #petroleumEngineering #machineLearning #DigitalTransformation

11

1 Comment

Awesome to see the Meta #Llama 3.2 models on #Azure #AI Model Catalog through managed compute inferencing #MaaS and soon on serverless API deployments while MACC Eligible providing customers with Enhanced Security and Compliance, Content Safety Integration, Simplified Assessment of LLM flows, Client integration, and Simplified Deployment and Inference. https://lnkd.in/g4Zpb3Dm

9

1 Comment

AI is transforming decision-making in the energy sector, helping leaders cut through complexity with clarity and precision. In Credera ’s blog, Nate Raymond shares how AI is enabling energy leaders to overcome decision distress and optimize their choices with confidence: https://lnkd.in/gMQ9s36t #Energy #AI #ArtificialIntelligence

4

Texas is the State of the Future. Learn more at the YTexas Summit. https://lnkd.in/gCiDhbzk Dozens of Intelligent Infrastructure Economic Zones are under development. Providing the NVIDIA-enabled INTELLIGENT INFRASTRUCTURE powering solutions like Artificial Intelligence, Spatial Computing, Intelligent Transportation, and Autonomous Systems. The Federal Infrastructure Bank and infrastructure investors are mobilizing $billions in private capital to underwrite Intelligent Infrastructure Economic Zones, which is revitalizing infrastructure in the United States. https://lnkd.in/gWnJD877 Texas SH130 is the nation's Route 66 INTELLIGENT INFRASTRUCTURE. https://lnkd.in/e5iwTNPR Building the Infrastructure for the Next 100 Years. Autonomy Institute https://lnkd.in/gwv2pnXe #infrastructure  #future #supplychains #autonomous #autonomy #transportation  #arpa-i #arpax  #jobs  #bank  #digital  #innovation  #data  #people  #visionzero  #continuumofcapital  #dataexchange  #partnerships  #digitaltwin #privacy #investment #p3programs

84

3 Comments

Liquid cooling is a topic that scares a lot of folks that haven't had to think about foundational infrastructure before, and now MUST consider it with the rise of AI-related compute. This blog article co-written by my esteemed colleague Lindsay Schulz provides a rapid understanding of how to think about liquid cooling in the context of your datacenter provider- and how to prepare yourself to debunk any misinformation that comes your way. https://lnkd.in/eDvwmbDB #LiquidCooling #AI Equinix #GlobalPrincipalsRock

12

In the 5th installment of the 11-part series from Decoding ML on Building Your LLM Twin, you’ll learn how to implement four retrieval and post-retrieval advanced optimization techniques to improve the accuracy of your RAG retrieval step. Read the full series to learn how to use LLMs, vector databases, and #LLMOps best practices to design, train, and deploy a production-ready “LLM twin” of yourself!

21