Francesco Iorio

Use it. Build it. Break it. 🛠️ I sat down with Luka Lamaj and the Microsoft Reactor to talk about the AI landscape, Canadian startups and how larger organizations are navigating their own AI transformations. Hang on to the end to hear what quote guides my work daily 💡 https://lnkd.in/gspwd_NW #AI #CanadianStartups #Founders #Innovation #TechOdyssey

17

2 Comments

This presentation that I gave last week at an HDRN Canada meeting provides a summary of our work thus far on synthetic data generation. It reflects the lessons learned about how to apply synthetic data generation methods effectively and about what works and doesn't.

41

#Quantum computing is not science fiction anymore. Executives need to draw their attention to post-quantum cryptography and beyond. Join me at #qConnect2024 in Calgary to explore the future of #sustainable #communities. Ask me about this at IEEE #Quantum Week.

37

1 Comment

Transcendence Series: The Founders and Leaders of Deep Learning   Yoshua Bengio, Geoffrey Hinton, and Yann LeCun are three Canadian computer scientists who are widely regarded as the founders and leaders of deep learning, a branch of artificial intelligence that uses neural networks to learn from large amounts of data and perform complex tasks such as image recognition, natural language processing, speech synthesis, and self-driving cars. They have also received numerous awards and recognitions for their work, including the Turing Award, the highest honor in computer science, in 2018. Yoshua Bengio is a professor at the University of Montreal and the founder and scientific director of Mila, the Quebec Artificial Intelligence Institute, the largest academic research center in the world dedicated to deep learning. He is also a co-founder and senior fellow of the Vector Institute, a research institute for artificial intelligence based in Toronto. He has published over 500 papers and several books on machine learning, deep learning, neural networks, and related topics, such as Learning Deep Architectures for AI, Deep Learning, and Neural Networks and Learning Machines. He is also an entrepreneur and a co-founder of several companies that apply deep learning to various domains, such as Element AI, Imagia, and Lyrebird. Geoffrey Hinton is a professor emeritus at the University of Toronto and was vice president and engineering fellow at Google - he departed on 2023 citing concerns on the usage of AI technology. He is also the chief scientific adviser of the Vector Institute and a founding director of the Gatsby Computational Neuroscience Unit at University College London. He has published over 200 papers and several books on machine learning, neural networks, cognitive science, and artificial intelligence, such as Parallel Distributed Processing, Connectionist Models of Cognition and Perception, and The Handbook of Brain Theory and Neural Networks. He is also an inventor and a co-inventor of several algorithms and techniques that are widely used in deep learning, such as backpropagation, Boltzmann machines, contrastive divergence, and dropout. Yann LeCun is a professor at New York University and the chief AI scientist at Facebook. He is also the founding director of the NYU Center for Data Science and the co-director of the NYU Center for Neural Science. He has published over 300 papers and several books on machine learning, computer vision, neural networks, and artificial intelligence, such as Convolutional Networks and Applications in Vision, Deep Learning, and Theoretical Computer Science. He is also an innovator and a pioneer of several breakthroughs in deep learning, such as convolutional neural networks, the LeNet architecture, and the MNIST database.   Note: This is the twentieth post in the series. The goal is to highlight significant figures who have made an impact on technology and its application to human endeavors. 

4

ARE WE IN A RECESSION? As mentioned in our prevision post, the NBER's Business Cycle Dating Committee took 4 to 21 months to declare past recessions. Sometimes, they even do so AFTER the recession ended. Therefore, how do we know whether we are in a recession today? And are we in one? We know by looking at a handful of indicators. One is the Smoothed U.S. Recession Probability (at the top on the image below), the other is the Real-time Sahm Rule Recession Indicator (at the bottom). Both datasets are updated monthly. The former tells us that we are not in a recession, but the latter suggests that we might be. Because of this uncomfortable ambiguity, the STOIC Recession Index will include a real-time horizon, which will provide the probability that we are in a recession. It will be updated on a monthly basis as well. Right now, it would probably stand between 25% to 50%, meaning that we are not in a recession, but that we might get very close to one.

4

🛑 🚫 Accountability in the Age of AI & Automation 💡💡 Air Canada came under further criticism for later attempting to distance itself from the error by claiming that the bot was “responsible for its own actions”! This case highlights a critical challenge as businesses increasingly adopt AI-powered tools: who is accountable when AI makes a mistake? In a first-of-its-kind ruling in Canada, Air Canada was held responsible for its chatbot’s erroneous advice. The company’s argument that the bot was a "separate legal entity" sparked both legal and ethical questions about the oversight of AI systems. Ultimately, the tribunal rightly emphasized: “A chatbot is just part of the company’s website. Responsibility lies with the business.” For professionals and leaders, this is a wake-up call. Whether you're deploying AI in customer service, healthcare, or finance, ethical design and accountability must be non-negotiable. What do you think? Are businesses moving too quickly with automation without ensuring the right safeguards? Let’s discuss! 👇 #AIethics #AI #Automation #Accountability #Leadership

32

3 Comments

Humans Beat AI: Taylor Swift Edition As the world’s brightest minds gather in Vancouver from December 9th to 15th for NeurIPS, the premier conference on neural information processing systems, one might think nothing could overshadow the advancements in AI and neural networks. Yet, the reality on the ground tells a different story. Enter Taylor Swift. While researchers prepare to discuss the future of AI, Vancouver’s hotel market has reached a fever pitch—thanks to Swift's Eras Tour stopping in the city on the same dates. The result? An unprecedented shortage of hotel rooms, with even the most modest accommodations going for upwards of $650 a night. The juxtaposition is almost poetic: on one side, we have the world’s leading experts in artificial intelligence, pushing the boundaries of what machines can do. On the other, one of the most influential pop stars of our time commanding such a massive audience that she’s effectively “disrupted” the plans of thousands of tech enthusiasts. It’s a reminder that, while technology might be the future, culture—powered by human passion—remains a dominant force. After all, it seems Taylor Swift is giving AI a run for its money… at least when it comes to securing a place to sleep. Humans beat AI? You be the judge. But one thing’s for sure: the intersection of culture and technology has never been more fascinating. #NeurIPS2024 #TaylorSwift #AI #TechVsCulture #Vancouver

2

We have 200 engineers and executives here in Toronto today for the inaugural The AI Summit Series The AI Summit at Security Education Conference Toronto (SecTor). 📊 Key Insights from the first panel - AI Governance & Security - a fascinating discussion on the intersection of AI, audit, and cybersecurity. Thank you to our speakers from Canadian Tire Corporation for bringing their expertise and Priya Mouli for guiding us through this session and the day as a whole. 🎯 Early Engagement Wins The most successful AI implementations involve audit and security teams from Day 1 - not after deployment. As one speaker noted: "If you're showing up two years later, after everything's been implemented... you've then missed the code." 🏗️ Build on What Works Organizations seeing success are building AI security on existing cybersecurity and governance frameworks. No need to reinvent the wheel - enhance what you have. 🤝 Cross-Functional Approach The future of responsible AI requires collaboration between legal, risk, compliance, and security teams. Siloed approaches won't cut it in this new landscape. 💭 Quote of the session: "We're now working with the most primitive interactions with AI we will ever work with... and that's always a staggering fact." #AI #Cybersecurity #TheAISummit #SecTor

58

4 Comments

I've just been reading fellow Torontonian Brett Kennedy's new book "Outlier Detection in Python" (https://lnkd.in/g_HgfnBq) from Manning Publications Co. Real world data is never clean and when doing data science and machine learning, you have to deal with that fact. Before deciding what to do with your outliers, you have to be able to detect them. Practical applications include financial fraud detection, network security and intrusion detection, bot detection, and manufacturing tolerances. Brett's book walks you through both the theory and the practice. You'll learn how to use Python and popular third-party libraries to detect outliers, as well as include the process as part of your data pipeline. The book is currently in early release, which means not only do you get great content, but you also get to participate in making the book better. Plus, MEAP is just fun to say. I'm forever grateful to my early readers, and now you can be one of Brett's. Definitely worth your time.

20

2 Comments

Canada's AI Dilemma - From Pioneers to Silicon Valley's Farm Team? Canada has long been at the forefront of AI research, boasting 10% of the world's top-tier AI researchers and significant government investments, including a $2.4 billion allocation in 2023. Despite this, the commercialization of AI remains a challenge. Prime Minister Trudeau acknowledges the gap, emphasizing the need to retain intellectual property and talent within Canada. Experts argue that while Canada excels in AI research, it often loses its innovations to foreign tech giants like Uber, Disney, and Nvidia. This has led to concerns that Canada is at risk of becoming the "farm team" of the AI world, where its brightest minds and innovations are harvested by international firms. Additionally, there's a fear that Canada could miss out on lucrative AI opportunities and become a non-player in the global AI market. Benjamin Bergen from the Council of Canadian Innovators highlighted the issue of AI intellectual property being institutionalized to benefit foreign companies, while Jim Hinton's research shows a significant portion of patents produced by Canadian researchers ending up in the hands of these giants. The government's efforts must focus on ensuring Canadian companies can effectively commercialize AI technologies and retain IP ownership. With the right strategy, Canada has the potential to become a global leader in responsible and democratic AI, rather than a secondary player in its own innovation landscape. Are you as concerned as me about Canada turning into the global AI farm team? We have slow and low AI adoption rates at the corporate level. We train industry experts just for them to go state side. Thoughts? #AI #Innovation #Leadership #TechCommercialization #CanadaAI** https://lnkd.in/gaaZr-CB

6

Despite Canada’s world-class institutions and talent producing remarkable AI research and products, commercialization remains a significant challenge. Canadian startups struggle to find customers, hindering their growth. The ecosystem excels in innovation but lacks the market support to transition from research to commercial success. Addressing this gap is crucial for leveraging Canada's AI potential fully. Read more about this issue here(https://lnkd.in/gjjWU4GZ).

14

Using RAG (Retrieval-Augmented Generation) with LLM can be a really powerful technique for improving LLM (or SLMs) response accuracy. This technique allows to provide data (based on documents, images, ...) to the LLMs that he can use to answer. This means that quality of response is related to the quality of data provided. here are some interesting articles to improve and secure your RAG projects ! Using Azure AI Search for RAG : https://lnkd.in/eAxKRugy chunking techniques : https://lnkd.in/ehzNTCuY multimodal RAG : https://lnkd.in/eHKxRYuD access control : https://lnkd.in/epBTHTDK for complex discovery : https://lnkd.in/ep5Cnn3p RAG for Structured Data Processing : https://lnkd.in/ePAF3mPk

4

Curious about combining AI with real-time data retrieval for more accurate results? Discover how to build Retrieval-Augmented Generation (RAG) applications on VS Code's AI Toolkit! https://lnkd.in/gVQPKwZi In this blog, Vinayak Hegde breaks down the process of creating a RAG database using ChromaDB, LangChain, and Python, transforming document retrieval into a seamless experience. From chunking documents to embedding vectors for precise searches, learn how this hybrid approach can enhance content generation, legal references, and more. Dive in and start building your AI today! #RAG #AIToolkit #LangChain #ChromaDB #msftadvocate

18

🚨 The AI Revolution Needs Transparency! 🚨 Did you know that generative AI systems can whip up text, images, videos, and even computer code based on just a simple prompt? 🤯 But there’s a catch… what's actually happening in that 'black box' of AI training is a mystery. Imagine trusting an autopilot without knowing how it can handle turbulence. This is what we're facing with AI. 🔍 Creators are now calling on Ottawa to demand disclosure on how these AI systems, like OpenAI's ChatGPT, are trained. Why? Because transparency is crucial for innovation that we can trust. Think about it: - 📈 Enabling better, more ethical AI development. - 👥 Building trust with consumers who use these AI-powered tools. - 💡 Fostering a more informed and involved tech community. This could reshape our digital future! Check out the full article for more insights: [Read here](https://lnkd.in/g_J-zPCe) What's your take on AI transparency? Let's discuss! ⤵️ #AI #TechTransparency #Innovation #EthicalAI

2

I just got my own copy of “Building LLMs for Production” by Louis-François Bouchard and Louie Peters. As I am getting more into applications of GenAI I expect this book to become one of my go-to resources in the upcoming months. https://amzn.to/3Yv5FmL #AI #GenAI #LLMs

2,251

51 Comments

Much has been said about many companies’ desire for more compute (and data) to train large foundation models. I think it’s under-appreciated that we also have nowhere near enough compute available for inference on foundation models. Years ago, when I was leading teams at Google, Baidu, and Stanford that focused on scaling up deep learning, many semiconductor makers, data center operators, and researchers asked me if AI would continue to make good use of more compute if they kept delivering it. For many desktop workloads, like running a web browser, a faster CPU doesn’t help much beyond a certain point. So do we really need faster and faster AI processors? Each time, I confidently replied “yes!” and encouraged them to keep scaling up compute. (Sometimes, I added half-jokingly that I had never met a machine learning engineer who felt like they had enough compute. 😀) Fortunately, this prediction has been right so far. However, beyond training, we are also far from exhausting the benefits of faster and higher volumes of inference. Today, a lot of LLM output is for human consumption. A human might read around 250 words per minute, which is around 6 tokens per second (250 words/min / (0.75 words/token) / (60 secs/min)). So it might seem there’s little value to generating tokens much faster than this. But in an agentic workflow, an LLM might be prompted repeatedly to reflect on and improve its output, use tools, plan and execute multiple steps, or implement multiple agents that collaborate. So, we might generate hundreds of thousands of tokens or more before showing any output to a user. This makes fast token generation very desirable and makes slower generation a bottleneck to taking better advantage of existing models. That’s why I’m excited about the work of companies like Groq, which can generate hundreds of tokens per second. Recently, SambaNova also showed it can hit hundreds of tokens per second. Incidentally, faster, cheaper token generation will also help make running evaluations (evals), which can be slow and expensive today since it involves iterating over many examples, more palatable. Fortunately, both training and inference are rapidly becoming cheaper. I spoke with Cathie Wood and Charles Roberts of the investment firm ARK, which is famous for its bullish predictions on tech. They estimate that AI training costs are falling 75% a year. If they are right, a foundation model that costs $100M to train this year might cost $25M to train next year. Further, they report that for “enterprise scale use cases, inference costs seem to be falling at an annual rate of ~86%, even faster than training costs.” I don’t know how accurate these specific predictions will turn out to be, but with progress in both semiconductors and algorithms, I do see training and inference costs falling rapidly. This will be good for application builders and help AI agentic workflows lift off! [Original text: https://lnkd.in/dJ9tVGh7 ]

2,900

80 Comments