Bob Frankston

During yesterday's LibreQoS APNIC webinar, I posed the question: won't more bandwidth solve these problems? As Herbert Wolverson said, yes, more bandwidth is good, but, still these problems remain if queueing is done incorrectly. Here's my take on why bandwidth alone is not as good as bandwidth + good queueing policies: Generally it was believed that 'data is important, so don't throw it away; hang on to it and send it later'. In practice, this has proven to be suboptimal as two issues may emerge: 1) latency increases for some flows due to heavy demand from other flows using the same bottleneck link, 2) even a single flow can have excessive latency due to aspects of typical TCP behavior (referred to bufferbloat) when the buffers grew large enough that the data being buffered was retransmitted anyway. It turns out that not all data is equally important. Active Queue Management is the art of deciding priorities, both in deciding what data to throw away, but also in allowing some later arriving data to be transmitted ahead of data in another connection that arrived before it. These problems have been studied, and good solutions have been found by using certain queueing policies in routers and switches, referred to as “fq_codel’ and “cake”. These track the different flows not by classifying the data, but by watching the behavior. Flows that send relatively little data (DNS lookups), or at a measured pace (video chat) have priority over flows that send a lot of data as quickly as possible (App and System updates, Video and ISO downloads).

10

5 Comments

Andrej Karthpathy, one of the world's foremost AI researchers announced his new AI ed-tech company, Eureka Labs. "This Teacher + AI symbiosis could run an entire curriculum of courses on a common platform. If we are successful, it will be easy for anyone to learn anything, expanding education in both reach (a large number of people learning something) and extent (any one person learning a large amount of subjects, beyond what may be possible today unassisted)." #edtech #frontiertechnology #startups https://lnkd.in/gSwGaBUB

8

Here's a list of the 28 books I read in the first half of this year. It was difficult to choose, but 5 specific recommendations are in my latest issue of #TheLongHorizon. (Here: https://lnkd.in/e2ftHs8H) If you are not yet a subscriber, here's a link to do so if interested: https://lnkd.in/em-5zEG5 What books did you read in 1H that you'd absolutely recommend?

10

2 Comments

Vinod's thoughts on AI: Key Takeaway 1: The impact of AI on various industries is expected to be highly deflationary. As AI automates many tasks and reduces labor costs, prices for goods and services are likely to decrease significantly across sectors like healthcare, content creation, and professional services. Key Takeaway 2: There are still opportunities for AI startups, particularly in niche applications and specialized domains. While large language models may become commoditized, there is potential for domain-specific AI solutions that integrate workflow knowledge and industry expertise. Key Takeaway 3: The cost of using AI models is expected to decrease dramatically in the near future due to competition between major players. This will likely make AI more accessible and ubiquitous, though training costs remain high. Key Takeaway 4: Apple's recent AI announcements demonstrate the company's ability to integrate technology, platforms, and products. This positions Apple well to bring small on-device AI models to consumers and developers at scale. Key Takeaway 5: The upcoming U.S. presidential election has significant implications for the tech industry and AI development. There are differing views in Silicon Valley on regulation and policy approaches, with some favoring less regulation while others are concerned about ethical issues. https://lnkd.in/gG9hSuUy

11

2 days ago California State Senate passed SB 1047, which establishes safety standards for AI developers of frontier models. A lot of the requirements directly demand guardrails that ZenGuard AI is developing. SB 1047 has far-reaching implications as it will inform other regulations in other states and countries. https://lnkd.in/dNSfJqun * All developers are strongly encouraged to follow safe, secure, and ethical AI development. The measures are (among others): pre-deployment safety testing, red-teaming, cybersecurity, safeguards to prevent the misuse of dangerous capabilities, and post-deployment monitoring. The goal is to set a community standard for safe, secure, and ethical AI development, that even smaller developers can adopt. * SB 1047 heavily targets its regulatory focus on "frontier" AI models requiring computing power greater than 10^26 floating-point operations that cost over $100 million to train. * The bill introduced CalCompute - a state-backed cloud computing cluster to democratize access to GPUs for smaller developers. In conjunction with the standards and guidelines in the bill, there is a clear motivation to adopt safety and security at the earliest stages of AI development for all developers. * This sets legal framework for attorneys to take legal action in the event the developer of a powerful AI harm or threat to the public safety.

17

4 Comments

Here's an interesting bit of research to keep in your thoughts: According to IEEE Spectrum, the very same Large Language Model (#LLM) that power generative AI platforms such as #ChatGPT and others are actually capable of passing, even outperforming humans in Theory of Mind tests. Before you panic, this doesn't necessarily mean that LLMs actually do possess theory of mind or self consciousness -- but it does pique interest in theoretical purposes and exercises. In reality, while LLMs are scoring well, there is no real method for how to test for the existence of theory of mind -- so again, #AI here is just doing what it does best: #synthesis of data from many different datasets. What this does mean is that AI systems can better mimic and understand human psyche and anticipate what they need, not necesarily that it can make decisions or form thoughts in itself. Is this breakthrough exciting to you? How do you think AI technology will improve with this one aspect?

2

2 Comments

This analysis, at least anecdotally makes sense to me. In many situations, infringe and holdout is the least painful option for large infringers. The system enables it. - The plaintiff must allocate the time and $$ to assert, taking time and $$ away from selling and developing products. Even if they engage litigation financiers they are limiting their upside potential. - The infringer continues to infringe while employing stall tactics or simply relying on the painfully slow, overburdened legal system. - The cost benefit analysis for infringers - "infringe and pay lawyers" typically outweighs the cost of design around or license. - Enhanced damages and injunctions are rare. - Litigation insurance products reduce risk all around. "The Marcum report shows that there is indeed something wrong with the patent system in the United States but it’s not excessive litigation or patent trolls. The real problem is the inability of patent owners to enforce their rights against infringers. Given the difficulty of obtaining injunctive relief and the limited enhancement of damages, it is unsurprising that predatory infringement has become a standard business practice when, at worst, all an infringer can be expected to pay is a reasonable royalty and, at best, nothing at all."

5

❓ How do we nurture students who can think and solve problems INVENTIVELY, ADAPTIVELY & CRITICALLY -- a Masterclass for Teachers Knowledge WAS Power. It no longer is. ❌ With Generative AI, some have argued that the end of search engines is near since knowledge can easily be obtained with just some good prompting. To survive and thrive in this 21st Century, where things change so rapidly, we need to use knowledge in a way that will help us solve problems --- this is the one competency that will stand the test of time and human development. We need to be able to use knowledge inventively, adaptively and critically. But it isn't so straight forward for educators and parents. After all, we were brought up in an age when Knowledge was power. Thus, as requested by fellow educators, I have spent the last few months putting together a masterclass for educators and the enlightened parent to help them facilitate students of all ages acquire competencies that enable them to do just that. 📣 Now I am happy to share this course with my friends in LinkedIn! It is broken into bite-sized video discussions of a few minutes each so the busy educator can explore as and when you have the time. Check it out at 👉🏼 https://lnkd.in/gmjcBy8y Use the discount code: PSC30OFF to get 30% off the course fees, limited to the first 100 people. Hope you find this useful! Let me know in the comments what else you like me to prepare and share with the community in the areas of education, teaching and pedagogy! #education #21CC #emerging21CC #problemsolving #inventivethinking #adaptability #school #pedagogy #masterclass #teachers #parents Eric Lam I help teachers promote Emerging 21st Century Competencies. I co-wrote works on next generation learning and the staple biology textbook 📘Biology Matters 📗 used throughout Singapore and the region. I invite you to connect with me and ring the 🔔 to see my latest post.

9

2 Comments

This is quite a piece on accusations of unethical pay-to-play within the #cyber #VC ecosystem. -Guaranteed first-year sales for portfolio companies -Formal loyalty program for CISOs -Points awarded within the program for CISO commitment level to portfolio companies (meetings, PoCs, product purchases) -Share of GP fees for CISOs in the loyalty program https://lnkd.in/gDta7Gp7

31

2 Comments

Some unicorns and late-stage VC-backed companies restrict their employees from selling their stock, often at the request of their boards of directors. Not this one. Andrew Feldman and his executive team at Cerebras Systems (a high-flying AI chip startup mounting a challenge to the likes of NVIDIA and Groq) have been willing to recognize the blood, sweat and tears that employees with vested options have put into building this impressive business, by allowing them to sell their shares in the private market. This is not just great news for current and former Cerebras employees but also for the company and its investors! Why? Liquidity typically leads to higher stock pricing, making it more attractive to investors. How so? The short answer is that when a stock is more liquid, its riskiness declines. When risk goes down, so does the discount rate (the cost of capital) that investors use to price the asset. When the discount rate goes down, all else equal, the price should go up! This enhanced liquidity and pricing can help a company with fundraising valuations, consolidating the cap table before an IPO, and bridging the valuation gap between an illiquid private market and a very liquid public market. Congratulations to Andrew and the Cerebras team for delivering on your promise to your employees and investors by allowing a liquid market for your stock. And bravo to your board members (which, according to PitchBook include Eric Vishria of Benchmark , Brad Gerstner of Altimeter, Steve Vassallo of Foundation Capital, Lior Susan of Eclipse, Thomas Laffont of Coatue, and Pierre Lamond) for supporting your vision! #secondarymarket #privatemarket #marketplace #unicorns #stockoptions

89

7 Comments

Join C10 Labs and their AI Venture Studio for a roundtable discussion on building AI-first ventures. This is an opportunity to meet founders from the 1st cohort, network with other applicants from the 2nd cohort, and connect with early-stage entrepreneurs in the AI community. Don't miss out on the chance to learn from the experts and build your own AI venture. Register now! #AI #entrepreneurship #C10Labs #CIC

48

1 Comment

Thanks to Microsoft and Bowmans (Law Firm) for a great “Responsible AI” session today. Lots of talk on policy, tools, compute capacity, and of course talent & skilling. Made me realize we’re often conflating what are actually 4 distinct and wicked problem statements: (1) At a labor market level, how do we avoid large-scale unemployment and dislocation as Gen AI threatens a huge range of jobs (particularly junior digital jobs)? I find this particularly scary as the digital labor market is flat and borderless, and there’s a risk that many of the most entry-level digital jobs (the ones the rising generation of African youth could fill) will be wiped out by Gen AI imminently. (2) At a broad-based professional level, how can everyone do their jobs better with AI? The reality is that all of us, across careers, sectors and seniority, can and should be making use of Gen AI to boost productivity, use data better, and make ourselves obsolete. This isn’t confined to techies and programmers; this is relevent to pretty much everyone in broad “knowledge work” positions across ops, HR, finance, consulting, and more. (3) At an African youth level, how can young people “get a job in AI”, and what does that even mean? This is more about understanding and developing the specific career pathways into the AI universe, which we’re just starting to understand. This includes engineering of course, but also a broad range of other job types around data annotation and cleaning, prompt engineering, automation engineering, and much more. (4) At a social impact level, how can we design more AI-based applications and use cases to drive more value and impact to those who need it? AI is poised to disrupt and transform a broad range of how the world works day to day, and hopefully many of those applications and use cases make life easier for those most in need — applications around health, energy access, food access, livelihoods and income generation and more. (In fact, Shortlist Futures / Shortlist Professionals is about to launch a new fund looking at positive use cases for Gen AI for women’s economic empowerment, job access and income generation - watch this space!) While there are certainly many areas of overlap between these problems, each needs its own attention, research, and creative solutions. Others in the room (Winnie Karanu, Martin Ndlovu, Isis Nyong'o Madison, Rositsa Zaimova and many more) — what did I miss?

64

6 Comments

The magic of AI will only be unlocked for the agtech ecosystem once we have good ag data to feed the beast. We need digitized, harmonized, standardized, interoperable, machine-readable data for the magical AI tools to do meaningful work in agriculture. Let's get to work!

17

5 Comments

We are delighted to present the concluding installment based on our paper, "Vision 2035: The Emergent Singularity" (link in the comments). In this final section, we illuminate the three critical components that will prove instrumental for companies as they navigate the transformative landscape of machine intelligence: 1. The Role of Data: In the age of machine intelligence, data reigns supreme. As we navigate this data-rich landscape, challenges emerge in harnessing its full potential. The concept of "Big Data" evolves rapidly, demanding advancements in edge computing, natural language processing (NLP), and even quantum computing. Despite this growth, there's a growing reluctance to share data, hindering accessibility. Startups can serve as pivotal players, breaking down barriers and showcasing the value of innovative solutions. From technical to regulatory hurdles, the path to data accessibility is fraught with challenges, but also ripe with opportunities for venture-backed initiatives to lead the way. 2. Open-stack technologies: Rooted in openness across software, data, storage, and computation, open-stack technologies are analogous to open-source software in many ways. By embracing open-stack, companies transcend closed system limitations, fostering innovation and enabling agile, cost-effective development through collaborative ecosystems. Leveraging existing communities potentially reduces customer acquisition costs while cultivating dedicated user bases for lasting relationships. 3. Collective intelligence: Ecosystems built on shared, community, and aggregate knowledge offer prime opportunities for open-stack-centric machine intelligence startups. The emergent benefits from group knowledge can speed up product development and broaden customer reach. Collective intelligence approaches like federated learning can also streamline data processes, minimize data sharing while maintaining control and adaptability, enhance model accuracy, and mitigate technical and product risks across an ecosystem. We hope you have found our series of summary analysis on the realm of machine intelligence to be insightful. As highlighted in our longer vision paper, we are convinced that the value creation potential of winners in this space will amount to a remarkable two orders of magnitude (100x) compared to today's companies. Lastly, we may follow up with a little encore in two weeks. Keep your eyes peeled! Patrick & Sri Machine Ventures #singularity #machineintelligence #ai

12

3 Comments

Science-fiction is inherently speculative. Author George #Orwell of ‘#1984’ predicted concepts such as doublethink, little earbuds and society’s obsession with the black mirror screens that dominate our lives — and according to the BBC, novelist and short story author JG Ballard, of “High-Rise” and “Crash” fame, has predicted the current tool and cultural fixation of #GenerativeAI in the modern society — only he didn’t just speculate them within a story, he is actually one of the first use-cases of a generative AI used in a creative context — Ballard was so fascinated with the growing technological advancement of computers that he used one to compose poems, essentially the precursor to what we have now with ChatGPT. Today, the debate about the creative use of generative AI and its value on creative industries is an extremely fraught and controversial topic, but if controlled in a space that doesn’t replace human labor or try to equate itself to traditional art forms, it is interesting to see what computers can do and how they are creatively limited in comparison. Ballard’s two poems, The Yellow Back Novels and Machine Gun City are generated, and both include notes on how Ballard achieved producing them. Regardless of quality (Ballard himself probably wouldn’t describe these two poems as his best work), it is interesting to see primitive versions of the technology that we utilize on the daily today, and how these moral conversations that concern them would continuously evolve.

1

1 Comment

I find it hard to believe that with millions of EVs entering the market and the proliferation of AI compute that PGE is actually going to have excess capacity in 2040 without any additional generation from 3rd parties. I am also curious if there are enough transmission lines and if PGE views that as someone else’s problem. Maybe they are planning on AI and ACTUAL figuring it all out! https://lnkd.in/gybsMmaw

10

5 Comments

Where should you get started with GenAI? Russ Somers’ insights have really stuck with me. His credibility implementing AI is unquestionable — he’s tripled his department’s effectiveness through cleverly deploying a “GPTeam” of collaborators is quite an accomplishment, for sure. (Link to podcast below) But what really struck me most were his insights on the personal aspects of getting started. I’ve argued for the importance of using AI for oneself, but Russ shed fresh light on the question of where to start. Rather than diving straight into work applications, Russ emphasized the importance of play, discovery, and starting from a place of personal intrigue. This approach is often overlooked, to the detriment of the explorer, and the organization. Many organizations are asking their people, “What can GenAI do for our business?” This is the wrong question! Not because GenAI is irrelevant to the business, but because most employees don’t even know what GenAI can do, period. The necessary first step is to open their minds to the power of the technology, in a way that sparks their imaginations. One of Russ's key insights was the importance of avoiding complacency in building confidence in the power of these tools. If a suboptimal outcome is acceptable for your chosen point of focus, it shouldn’t be a candidate for GPT-powered exploration. True innovation demands iteration and refinement borne of a restless longing for more. With routine work, it’s sometimes easy to get comfortable and settle into the groove, but that’s not where significant gains will be made. Because innovation is accompanied by many false starts and frustrating dead ends, I’ve often told teams this sense of mattering is the essential prerequisite to undertaking a new project — if you don’t care, don’t bother. For whatever reason — unless you’re in a start-up environment with tightly aligned incentives — folks tend to demand more excellence from solutions to personal concerns than to work-imposed ones. To discover the value of GenAI, start personal. Having observed countless professionals attempt to incorporate this technology into their lives, I can tell you that your biggest barrier is likely going to be summoning your own activation energy. You’ve got to find a project where you’ll be willing to push beyond mediocre outcomes. Starting with a personal project is an exceptional hack to invoke care.

24

8 Comments

Gift link for op-ed from the Editorial Board of the WSJ. We're in much agreement on this. From my perspective, more federal R&D should have been funneled towards safe and responsible AI all along, rather than reacting to winters and springs (consistency is key). Canada maintained a small investment through the last AI winter and it paid off nicely. It's one of the reasons they punch above their weight in AI today. We maintained our research efforts in KYield as well in a very lean manner, which is why I think we are still well ahead of the pack in our focus areas. But the idea that we need to suddenly invest $32 billion per year in non-mil AI is primarily buying votes in academic research across the country. Big Tech spends $220 billion a year in R&D. The primary problem is most of it is strategic to Big Tech, not customers or society. Monopolies want to force us to spend more GDP in their products -- they don't particularly care how, nor are they incentivized to care. However, as the piece suggests, spending has increased in all industries -- it's not at all just about Big Tech, despite the headlines. ~~~~~ If China is “going to invest $50 billion, and we’re going to invest in nothing, they’ll inevitably get ahead of us,” Mr. Schumer said Wednesday in unveiling a “roadmap” for AI policy with Republicans Mike Rounds and Todd Young and Democrat Martin Heinrich. Invest in nothing? Didn’t Washington pass massive climate, infrastructure and semi-conductor spending bills? "Mr. Schumer used the same justification in 2022 to pass the $280 billion Chips and Science Act, though China trailed the U.S. by years in semiconductors. Now President Biden is doling out money to chip makers at campaign stops, including $19.5 billion this spring for Intel at a chip plant in Arizona." "The Senate AI report calls for Congress to pass $32 billion a year in “emergency” spending for non-defense AI. What’s the emergency? Perhaps the November election."

4

2 Comments