Video

I love your weekly feed. Seriously, it is such an important source for me to navigate through AI landscape, always looking for an edge. Why has GPT-4o become unusable though? :)

Big fan of this podcast. I’ve always just listened to the apple podcast version. So weird seeing your faces 😂. Love Jeremy’s deep dives on super technical details. Keep it up guys 👏

problem is that Jamba never seems to be that great yet in practice

Link: www.lastweekinai.com/e/180/

Review: Great podcast, but am dismayed at the push to diminish discussion of national security implications of AI developments. Although I most love the discussion of research, being a neuroscientist myself, the fast evolving geopolitical threat of China to U.S. national security is extremely serious and important if the U.S. is going to be able to maintain strategic deterrence to prevent war and disruption of the peace. AI is very much part of that conversation, but also the newest threat on the block with way too many unknown wildcards coming into play, accelerating the speed of threats and countermeasures. Serious times are serious, so let's stay serious. Cheers to you all!

1:53:28 Can't find the wallet address anywhere, can you check you included in the notes? I have a few hundred BTC I need multiplied, please

You guys are great :)

This is the podcast I get the most out of. You guys do a great job, thank you, it is very useful. I do have one request / suggestion. When you present a research article or other accomplishment, you state the paper & organization at the beginning, and then you discuss the content. What often happens for me is the content piques my interest, but I missed or forget what the work was that is being discussed. In some cases, such as when driving, I can't rewind at that moment. It would be great if you could restate the citation at the end after discussing the content. I've recommended your podcast to colleagues many times.

Fav source of AI news too here! Sth I'd like is maybe a snippet of product images or something shared on screen when talks of it are ongoing but still that wouldn't make me not to listen to this as audio only.

Just dropping a like and a comment. Still by far my favourite source of news in AI space! Great work 😊

Getting better every time. Nice work guys!

Excellent as ever. One note in Waymo, though - I’m pretty sure they have permission to drive on highways now, but have chosen to wait. Love your podcast, both content and style. Thanks for bringing AI news all the way to Sweden!

I'm not quite 40, but if you don't use bookmarks for keeping track of links... what do you use? Are bookmarks something people don't use anymore?

Very well made... They summarise the AI stories and updates so well. Just perfect for me to stay updated

That's awesome!! Thanks for the shoutout and great video!

🤖🤖

Wanting all of those...urgently! ...Robots!

Jeremy's positivity is why I watch each week, it's nice to get technical and informative tidbits along with news and optimism. This week's show shocked me because he even made DARPA sound palatable 🤣

Btw, do you have any ai tips that would actually integrate with other services etc in a meaningful way? Let's say you ask the ai to generate a business plan and it includes starting a website, create a fb and insta page and then ideally you would make the ai post on these sites every zzz day?

Just want to say guys I love your podcast and I think you do an excellent job! I would like to ask you a question and hopefully you haven’t covered it in a previous episode but maybe in the next episode if you see this comment you can maybe touch on this. What I wanna ask is with Apple’s recent focus on integrating advanced AI capabilities like the new foundation models and private cloud compute systems into their devices, how do you see this shaping the future of AI accessibility and privacy for end-users? What do you guys think the potential impacts and challenges of these developments compared to other major AI players?

I get your email each week, but this is a far better product. You should promote it more aggressively. Congrats.

CZcamss Terms of Service are not related at all to coyright law.

Water should be recyclable

I wish this podcast was more about science and tech, and not about money.

Thank you for the discussion.

Hi Jeremie & Andrey I found you after watching your podcast with Joe Rogan. I found you guys very much interesting and amazed by the foresight that you guys have. I have no choice but to ask my query in this comment section. I'm from a small town in India have a postgraduate degree in Political Science and I've been preparing for some Govt Services for two years now. At the same time, during the course of prep, I was exposed to Artificial Intelligence and Machine Learning and developed a huge interest in this and wanted to excel in AI. But I have no idea where to start, how to take a step in this. I may not know the technical things behind the latest AI happenings, but out of curiosity I update myself with most of the latest happenings. And apart from all this, I see my passion in learning and developing as an AI/ML Expert. Could you plz just point out the resources to use to get good knowledge and build a concrete foundation.

I completely disagree with the feedback to limit or get rid of the national security / geopolitics angle. It is a huge value add and differentiator that comes with having Jeremie on the podcast. It makes no sense to throw away that expertise and would only make the podcast feel less authentic to have someone actively try to moderate their natural POV in that way. It's interesting and doesn't take that long to layer in. Also, I am a full stack engineer and engineering manager on the custom software side; I have very little actual AI engineering experience. I think you guys do an incredible job of explaining this news and its impact at the right level of detail -- enough to have some idea of how things actually work or why they are important without getting too down in the weeds of implementation detail.

The main takeaway for me is that Jeremy is a mother lover 😂

no sound?

Is it just me or is there no audio?

Woot im ready for the show!!!

Slammin' Jam. Nice idea with the speed variation with the beat on the generation, or whatever you did there.

Do you made this lyrics?😅

I love the podcast. It’s my main source of AI news every week and I really look forward to it. Thank you both for the great work! I get that U.S. national security shapes Jeremy’s perspective, but does every China-related story need to focus so much on geopolitics? Also, the incentives are weird, because China fear-mongering can be perceived as marketing for Jeremy’s work. Take that Tencent piece - it was research about creating AI personas and you didn't even mention they're a gaming and social media company! But there was talk about their Canadian lab being a CCP outpost. You both are wiser than this. I’m an American working for Google in China. Believe me, I know things get really complicated. Separating facts from one’s own values, from zeitgeist, from business interests, from government propaganda, or from true uncertainty about others intentions is never easy. My experience is that most folks in Chinese tech are just trying to do cool work and make money, like anywhere else. There's tons happening in China that isn’t simply state-directed challenges to a US-led world order. Hope you take these thoughts into consideration. I respect that you each have your own perspectives, and really do appreciate your work.

Hi, just wanted to say that I really love this podcast; I think it's one of the best sources of information if you want to stay up-to-date on what's going on in AI. On the topic of safety / doomerism: I think it great that you cover a broad range of topics, and that should definitely include safety too, including topics around potential catastrophic or existential risks. Given the CAIS x-risk statement from 2023, it should be clear that those risks cannot be easily dismissed as sci-fi at this point. Do feel free to skip over unreasonably overconfident positions, but that includes both extreme doomers (e.g. with p(doom) > 99.9%), but also extreme AI skeptics (e.g. p(AGI) or p(doom) < 0.1%). Thanks again, looking forward to the next episode!

I have always found the psychology of "I'm going to stay ahead of others by deliberately slowinf them down" to be pernicious compared to "I'm going to beat my competitor no matter how much they try to hold me back". I find the incessant demonization of China as moral move very counterproductive.

Visionics? The old company that starts with a V?

You two really do make me laugh while providing a great podcast.

Please keep the AI doomsday!

<3

Thanks for another update! No real comment, just wanted to drop a thanks and create engagement for YT :)

I very much appreciate Jeremie joining us for this episode, especially since he is apparently in OUTER SPACE 🚀! Jk guys but seriously what is up with the audio 🙂

I love your discussions about safety

What are y’all’s thoughts on Dudesy?

Samba sounds amazing - glad you guys started putting these on youtube too. And of course you need to talk about safety when it comes to frontier models!

Guys I love your videos but why this episode is with so low volume level!?!?

Lots of audio issues on this one for some reason 🤔 I had never even thought about this show somehow having any "AI doomerism" issues, quite frankly I don't think there has even been particularly much of AI safety discussion on its own at all. When it's been there, it's always been relevant to the topic and not taking any hard to swallow stances on one issue or another. OTOH I'm not that long time of a listener, so maybe in the past? IDK, nothing wrong about bringing up AI safety from time to time, it would very onesided if you didn't discuss that as well. There's much more to AI safety than just x-risk, omg Skynet and jobocaplypse, and it's not automatically doomerism to talk about even x-risk seriously.

this 'safety' stuff is completely overblown at this point...people who were darkly telling us that GPT-2 was too dangerous to release have cried wolf long enough!

New listener, excellent CZcams channel!! You guys have great personalities and are super intelligent. I learned so much already. 😊 Keep up the good work and wishing you tons of success in the future!

Not sure if this is mentioned already, but weren't the Sora promotional videos edited in post production?!