Marek Michalowski

A great side effect of choosing to build in generative AI, specifically at the intersection of motion, physics, and game-engines, is that my social feeds are inundated with the latest research, demos, and products in these fields. When combined with the other half of my feed—human creativity in film, art, games, and VR/AR—it's awe-inspiring to contemplate the types of learning and storytelling we'll collectively experience in the coming years. Also, I'm rebranding my time spent looking into the future via my feed: #BLOOMScrolling

14

🧠 AI will be pervasive in manufacturing 🏭 whether we accept it or not. It's time to embrace the technology and drive meaningful value with its usage. 💡 If you are starting this journey or have tried but have yet to quite succeed in generating value, this blog is for you.

8

1 Comment

China is set to release the Astribot S1 which is a humanoid robot built for home chores. Robots like these could help with the caregiver shortage both at home and in large senior livings. So many in the US go un thanked and unpaid while taking care of their elders and children and still working jobs to survive. Robots like this when they be come cost effective could substantially reduce the burden of aging on all. https://lnkd.in/eTx5skx8

4

Isnt the point of LLMs (ai) is that we can now freely speak to systems with our own natural language. Free of hardware, or learning new processes. Why do we need more hardware to interact with Ai. Wouldnt the natural way to adopt is through our messaging applications we already use (today) to communicate with other humans using natural language. (they seem to be designed for that kinda thing out of the box..) 🤔 https://lnkd.in/gu6y6nef

1

5 Comments

There’s been speculation we’re approaching a trough of disillusionment for generative AI. That seems pessimistic. Here’s why: - There ARE amazingly helpful use cases - Adoption HAS been rapid - There’s ALWAYS a period of time spent figuring out the application layer for technology, this time is no different - Teams are already figuring out how generative AI can be helpful, interesting, and exciting In my own life my family writes songs for our birthdays with Suno. We create books that show new worlds with Firefly. Both produce amazing moments that wouldn’t happen otherwise. I run many things (including this post) through ChatGPT. Generative AI makes it easier to analyze external data, videos, and reports. Mostly, I ignore what it says, but sometimes it’s really helpful. Perplexity is a starting point for research into topics and a source for recipes while Gemini makes exploring interviews faster and more interesting. Marvin, Great Question, and UserTesting are all working on this too. I’m also excited to try new tools from Apple, Every Inc., and others. There are also so many new things to come from agents. To understand what's happening, we analyzed data about how designers and developers are building and improving products that include generative AI. The blog in the comments covers the moment we’re in and the path forward for teams. The conclusion: “Will AI live up to the hype? Not all technologies—even the most hyped ones—are truly transformative, at least not in the ways we might predict. The ones that are have something in common: teams who see the potential and are willing to experiment until they get it right. This kind of work is underappreciated, and it often requires pushing against disillusionment, embracing failure, and knowing when to put your head down and keep building. But on the other side, it might just be worth it.” #AI #design #digitalproduct

30

5 Comments

I continue to be alarmed at the progress of proposed California regulation SB 1047 and the attack it represents on open source and more broadly on AI innovation. As I wrote previously, this proposed law makes a fundamental mistake of regulating AI technology instead of AI applications, and thus would fail to make AI meaningfully safer. I’d like to explain why the specific mechanisms of SB 1047 are so pernicious to open source. To be clear, there are routes that regulators should pursue to improve safety. For example, I would welcome outlawing nonconsensual deepfake pornography, standardizing watermarking and fingerprinting to identify generated content, and investing more in red teaming and other safety research. Unfortunately, the proposed bill pursues a less beneficial and more harmful path. SB 1047’s purported goal is to ensure safety of AI models. It puts in place complex reporting requirements for developers who fine-tune models or develop models that cost more than $100 million to train. It is a vague, ambiguous law that imposes significant penalties for violations, creating a huge gray zone in which developers can’t be sure how to avoid breaking the law. This will paralyze many teams. You can read the latest draft of the law online. I’ve read through it carefully, and I find it ambiguous and very hard to follow. Developers who try to navigate the law’s complex requirements face what feels like a huge personal risk. It requires that developers submit, under penalty of perjury, a certification of compliance with the requirements of the law. But when the requirements are complex, hard to understand, and can even shift according to the whims of an unelected body (more on this below), how do we ensure we are in compliance? For example, the certification must include many different sections. One is an analysis of “the nature and magnitude of critical harms … the model might reasonably cause or enable.” But given that even leading AI researchers aren’t sure what harms models might cause or enable, how is a team of developers supposed to figure this out and declare — under penalty of perjury — that they meet this requirement? Further, some developers will be required to implement “protections to prevent … misuse of, or unsafe post-training modifications of, the covered model and all covered model derivatives … that are appropriate in light of the risks associated with the covered model, including from advanced persistent threats or other sophisticated actors.” Even leading AI researchers don’t agree on how best to “protect” AI models against these supposed risks, or what would be “appropriate.” So how are developers supposed to figure out how to comply with this requirement? [Reached LinkedIn length limit. Rest of text here: https://lnkd.in/g_NhQpni ]

2,722

203 Comments