Giuseppe M.

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

Coding with Claude 3.5 Sonnet is ~10x more efficient than GPT-4o or any other LLMs out there. The new Artifacts feature enhances user experience by generating and running code next to your chat. Wild - Try it for free: claude.ai #ai #anthropic #claudeai #llms #coding

6

1 Comment

🍓 How many ‘R’s are there in ‘Strawberry?’ I suppose it wouldn’t be Friday without yet ANOTHER major AI Development - Nvidia quietly launched their own version of an AI LLM. No fanfare, no press confences, no ringing of the bell at the NYSE. Just a covert release of an LLM Model that bests ChatGPT, Clause, Gemini, and a host of others in almost every benchmark. And yes, it can correctly answer how many ‘R’s’ there are in ‘Strawberry.’ (Which you would think is the simplest of questions but one that bedevils many LLMs).❓ Impact/Perspective: As more and more businesses begin to incorporate AI Technology into their Workflows and Strategic Planning, choosing the best LLM for your business is becoming increasingly critical. Make sure you understand the nuances between LLMs and how they apply to your business needs. If you’re seeking assistance in making sense of this fast developing subject, please reach out: Www.PVAIConsulting.Com #AIMarketing #AIBusiness #AITechnology https://lnkd.in/gMhTEqxs

4

California Governor Newsom vetoes the controversial AI Safety Bill! And the reasons behind it are worth noting. Stifling innovation is a major concern but so is the lack of nuance in the assessment of risk He states that the bill would 'give the public a false sense of security about controlling this fast-moving technology' Which honestly- is spot on! The Bill tries but both goes too far & not far enough. We covered the topic in great detail in our inaugural episode of TheOutpost.ai podcast Do check out our short of it here & let us know how you think the world will react to the veto! https://lnkd.in/gfs3vVN4

22

Was talking to some clients this week and something I asked them really caught their attention. Go ask the #AI vendor you're working with this question: "What data did you use to train your model?" We aren't far from a time where culpability gets even more acquainted with the AI running your business. You can see this in local NYC laws & the EU AI Act. Sure many have a long way to go (#Canada AIDA and #US Exec Order on AI) but you can see where we are heading. BTW... here are the answers you will get, I will let you decide which one is right if you're using AI for #business (A) It's none of your f'ing business bros. (I actually think they say f'ing... but if not, the attitude says it) (B) We're not quite sure what's all in it, but we took it from these places. Don't worry (implied) about the 200,000,000 lines with a copyright symbol on it ... oh and... ya there are a lot of Swastiska references in there, and yes we know they all aren't historical references... (C) Here is the data. Here is where it came from. Here are all the things we did to the data (like strip out copyright stuff). Here is a list of places we will not take data from because of hate or copyright or veracity issues. Here is the person that owns the data pipeline. Here is the group (including privacy, legal, engineering) that signed off on this data set. Any other questions because I can answer them all.

66

GPT-4o is getting further integrated into Windows, and OpenAI is reportedly also bringing it or something similar to Apple devices, with an announcement possible at Apple’s upcoming WWDC. From data I found online, this would give OpenAI’s models a footprint on up to 52.4% of devices, and Google’s models most of the rest on Android devices. (Though not all devices will be able to run updated operating systems with the models, however.) If natural language, including voice, becomes the dominant means of interacting with devices, then Google’s chokehold on search might not matter on non-Google devices. If you ask your Windows, Mac, or iOS device to answer a question, it will either just answer it, or search with Bing. Now, Google has a massive search partnership with Apple that could prevent that on iOS devices, but that partnership may also be affected by antitrust actions. So it will be very interesting to see what Apple announces at WWDC.

6

1 Comment

University of California, San Francisco Chancellor, Sam Hawgood, in an interview with Matthew Lungren MD MPH at #ConvergeSummit has outlined the use of #AI scribes to create clinical notes as well as other #AI-driven technologies. Key to their success: a group of practicing physicians with knowledge of AI ensuring that the technology fits into the physicians’ existing workflow, which improves physician adoption of the technology. Another key: measure success through reduction in physician burnout and improvement in patient satisfaction, not increases in physician productivity. Final key: have a real-time, frequent review of implemented technologies to ensure that they still support and are relevant to changing clinical care standards - and shut the technologies down if they are not supporting their purpose.

10

Recently I've watch the WWDC...And, let's talk about Apple Intelligence. Just using Tim Cook words and mixing with some thoughts about their new product rationale. ✨Tim :"It has to be intuitive and easy to use." ✨Tim: "It has to be deeply integrated into your product experiences. " 🤓 Me: Diggin a bit on this, many people assumed that Apple would just call OpenAI for all AI tasks. Instead, at WWDC, they revealed their own: ➜On-device model for simple tasks such as writing and summarizing content, including an on-device semantic index enabling Siri to take actions across apps. This means offline AI, many users will opt for Apple's on-device model for simple AI queries, avoiding public LLMs. ➜Private cloud model for advanced AI queries, ensuring your data remains secure. This means AI integrated across apps, OS, private cloud, and Apple devices ✨Tim: "Most importantly, it needs to understand you and be deeply rooted in your personal context, including your daily routine, relationships, communications, and other aspects of your life." ✨Tim: "And, of course, it has to be built with privacy from the ground up. " 🤓 Me: Unlike other AI players, Apple’s business model doesn’t require monetizing your data. Put simply, people don’t trust any other AI player with their data: ➜Google and Meta make money from targeted ads. ➜OpenAI has a lot of drama. ➜Microsoft is dealing with the fallout from its Recall feature. In contrast, Apple has long emphasized privacy. People’s trust in Apple to protect their data is a huge strategic advantage, as always I have my doubts but, let's see what happens.

3

inception llms: experimenting with a new LLM design pattern where I prompt Gemini to write code on my behalf which, in turn, uses Gemini to perform tasks - So meta, isn't it? One prompt for one task. in less than two hours created a full pipeline which would have taken days otherwise. Gemini wrote more than 97% of my code, discoverying libraries I was not aware of - overall, saving me tons of time as swe. Where do you think you can use this power in your product? try it yourself next time you develop a feature. I named it "inception llms"

76

10 Comments

Following-up on my previous AI posts, few decades ago I was hopeful about Tim Berners-Lee's vision of a semantic web where searches could be more meaningful making more sense of widely available WWW content beyond ad-driven search results. Unfortunately, that vision never fully materialized due to multiple challenges and effort needed to better structure and standardize content but I'm hoping that current Gen AI capabilities can get us closer to that dream. Nowadays, you can now see changes on Google Search behavior powered by Gen AI. I found an article by Hank Graber that summarizes this well: 1. Background - In the early days of the web, content was primarily focused on visual aesthetics, mimicking print layouts. - However, the web is dynamic, unlike static printed pages. - To harness the web’s power, content needed to break free from page-centric models and shift toward meaning-based structures. 2. Semantic Web Vision - Tim Berners-Lee envisioned a “Semantic Web” where machines analyze web data, transactions, and content. - This vision, proposed over 20 years ago, aims to enable intelligent agents that handle everyday tasks. 3. Recent Developments - Headless CMS technology and structured content models liberated content from page constraints. - Content is now organized based on meaning, independent of presentation. - Advanced AI language models like ChatGPT have further accelerated this transformation. 4. Semantic Web Benefits - Immediate Access: The Semantic Web grants machines human-like understanding of content. - Structured Content: Rich metadata allows machines to grasp context and usage. - Connected Web: It facilitates innovative connections across websites and data sources. In summary, the Semantic Web, coupled with AI advancements, promises a more efficient, interconnected, and meaningful digital experience. Read the full article here: https://lnkd.in/g9pak_uY

5

2 Comments

🚀 AI is the new steam engine - transforming industries, just like its predecessors. Now, harness that power with GitHub Spark! ✨ Build apps with just your words: - Dream it - Describe your app idea. - See it - Real-time previews as it builds. - Refine it - Iteration is easy. - Save it - Version control keeps your work safe. 🌐 Enhancements: - AI integration? Done. - External data? No problem. 🔍 The era of AI calls for: - Visionary Product Owners - Creative Designers - Innovative Architects #GitHubSpark #AIDevelopment #NoCode Documentation: https://lnkd.in/dBzHAutD

8

This, via the excellent Ethan Mollick, is interesting when it comes to comes to everyday business uses of AI. Its very hard to break out of the Google search type use of AI and then wonder when it doesn't do much. I think understanding Prompt Engineering and LLM basics are actually v useful if you have the time and inclination. But the rapid pace of model development, especially now in the reasoning space, means that some of that can now be internalised in the model. Jumping in deep and giving it as much info as possible, in natural language, and working with it can be much more fruitful.

6

Good read on CA SB 1047. One aspect of regulation is putting the burden on AI developers but not putting accountability on the adopters of AI. The regulatory and statutory world is very much evolving.

5

The cracks are showing—and this is just the beginning. This is as an early warning for industries everywhere: AI disruption is not a distant storm; it’s here, and it’s reshaping entire landscapes faster than ever before. The lesson couldn’t be clearer: evolve now, or risk becoming obsolete in the next couple of years. Bay Area tech company, down from $12B to $159M in value, lays off hundreds

245

21 Comments

SB 1047 is a pending California law that is essentially outlawing large open source models. If you like open source (and I do!) this would be very bad. And now it is reported that one of the advocates of the bill is trying to cash in on it 🤦‍♂️ The bill technically doesn't outlaw large open source models, but it requires "PWC type audits" costing "on the order of hundreds of thousands of dollars to single digit millions." As most open source authors or university research labs don't have that much money, it achieves the same outcome. It would also make AI start-ups vastly more difficult and expensive. The AI community (or at least the part of the AI community that likes open source) unsurprisingly are trying to stop it. Today, the article below reports that "Dan Hendrycks, an executive at Center for AI Safety — the firm whose lobbying arm helped draft the bill — co-founded a company called Gray Swan that offers AI safety compliance tools that seem positioned to provide the type of auditing data the bill would require". https://lnkd.in/gdutbeKq

64

14 Comments

Worth a read... "With past Llama models, Meta developed them for ourselves and then released them, but didn’t focus much on building a broader ecosystem. We’re taking a different approach with this release. We’re building teams internally to enable as many developers and partners as possible to use Llama, and we’re actively building partnerships so that more companies in the ecosystem can offer unique functionality to their customers as well. I believe the Llama 3.1 release will be an inflection point in the industry where most developers begin to primarily use open source, and I expect that approach to only grow from here. I hope you’ll join us on this journey to bring the benefits of AI to everyone in the world." https://lnkd.in/eRBBC9m5

3

Profound? Still digging in, but this story IMO shouldn't be about Wx predictions. The quote: "That’s very different from how traditional models like ENS work, which still rely on supercomputers to solve complex equations in order to simulate the physics of the atmosphere" .. is stunning. The model, without an understanding of the world in any real physics-based sense, is better than one that does.. ? Calls into question Yann LeCun's objective-driven approach to AGI?

4