ocanada_techguy

Summary UFD Tech recently posted yet another questionable strike by YouTube content moderation. However now the process is made worse by immediately rejecting first appeal and by eliminating the ability to even give reason for the appeal. My thoughts It seems YouTube is just knee-jerk obeying every corporate complaint? That they within seconds rejected appeal means they're not even considering them. Is it AI deciding? What far far worse however is they've removed the space to give reason for the appeal. What kind of appeal process is that? BS if you ask me. I am really hoping creators with clout, including LAN Show TechLinked LMG, rally together to express their disdain for these changes and if private persuasion doesn't work then public embarrassment. Sources

Yes you are right, and those are important distinctions, it doesn't "think" per-se. You yourself commented it's "not organic answer" and "there are filters put in place", so, what I am intimating is if there can be caveats and disclaimers there could also be, let's call them adjustments, or in the lingo "training" as to what to "say" (write, respond, opine, etc.) about itself. Remember, Microsoft Chatbot "Tay" on Twitter was turned into a racist a-hole loon in a matter of days, so it's very likely there are indeed controls on the results, the messaging, what would seem like it's assurances or opinions or conclusions could be tempered, "tuned", or even propaganda. Cheers

Last week Linus and Luke touched upon how retailers and tech retailers in the United States such as California have huge economies of scale, the sheer number of consumers allows super-specialized retailers to thrive whereas in Canada the retailers are shocked to discover the geography and population are so very different, (and even the spread-out geography makes online fulfillment a challenge). All true enough. It explains how Canadian Tire over it's life evolved quickly to be an "everything" store, the one-stop shop in town like the Hudson's Bay company back in the fur-trader days, or the Sears & Roebuck catalog. Also why London Drugs does electronics, they got in early and did it right building customer loyalty. The return of Zellers was mentioned, and Target's failure in that regard. If you don't know when HBC "closed" Zellers they "sold" many of the locations to Target swooping-in from USA. They learned the hard way in jurisdictions like Quebec where workers rights are well-protected (your mileage may vary, your own state/province might be ok treating employees like $#!^) turns out they could not simply fire/ "re-hire" the employees as if they're staring over again at zero, you buy the company you still owe them their position, seniority, pension, etc. But here's the thing. Because I know Luke Lafreniere has mentioned several times the team's great work on the inventory management system your team developed, this will likely be of deep interest to that team @LinusTech If you're like me, you'll enjoy this read immensely. Programmers know attention to detail matters, and it matters a ton. It's a great read and could easily be a case study in business school. Here is the fascinating story of how mundane and basic errors setting-up the crucial IT computer inventory software systems (information systems) led to the disaster that was Target Canada. THE LAST DAYS OF TARGET By Joe Castaldo https://archive.canadianbusiness.com/the-last-days-of-target-canada/

Summary ChatGPT voice gets interviewed for 24 minutes on YouTube about the implications of AI on Automation and careers. It's very interesting to hear what ChatGPT has been trained to think about itself and it's own implications. In this episode Intel Business' Darren Pulsipher Interviews ChatGPT from OpenAI about utilizing AI for automation, the ethics of using AI, and the replacement of information workers. Quotes My thoughts Considering the wild reactions of Linus and Luke about the new Bing with ChatGPT on the WAN Show right now tonight, this interview seems very relevant to me. Sources Automation with AI (ChatGPT) #122 | Embracing Digital Transformation | Intel Business - YouTube

Recently Patrick Kennedy of ServeTheHome did some explainers of Intel QuickAssist technologies. I found these extremely useful to put appropriate but much OVERLOOKED context to the recent reviews by LTT Anthony and other tech reviewers of the recent ADM Epyc announcements, but it also has me thinking about Intel Core vs Ryzen vs M2 vs arm et. al. So, Intel has built-in hardware accelerators for many tasks, which everyone seems to overlook. Audio playback, video playback, video/audio streaming/sharing, webcam use (see Intel IPU 6.0), and AI accelerators too, Neural Engine (GNA 3.0) Intel Deep Learning DP4a VNNI etc. and of course the Xe Graphics includes dual INT/Float bfloat etc "matrix maths" capabilities which OS and software can exploit for remarkable performance boost. Perhaps the classic from 2nd Gen onward is Intel QuickSync Video which was offering unheard of 13x-33x accelerations of CD DVD ripping audio/video encoding and transcoding. Over a decade ago they built-in hardware encode/decode so AES levels 128 256 for things like encrypting your storage (PC/phone/tablet) and httpS could be real-time and not strain your compute nor your battery. Apple has some also, they've recently touted a "neural engine" and added hardware support for ProRes (sidebar: curious how useful that is, as in what percent of consumers use it, is it fully in-place in the ecosystem). AMD as I understand it might finally get around to implementing AVX-512 on server side, they have "ML" in Ryzen called SenseMI as if it's ML but AFAICT (as far as I can tell) it's just turbo boost that claims to be smart about it? (sidebar: seriously, that's your AI??) Recently LTT team reviewed AMD Epyc with prognostications about how Intel would be trouble from this. Everyone seems to focus on how many cores. I tell you what though, if I can do considerably faster work with considerably fewer cores, to quote Mad Magazine Alfred E. Newman "What... me worry?" When reviewers benchmark games, sure there are complexities of bottleneck and a myriad of settings which can impact the outcome, most reviewers know to use "best drivers". However when benchmarking video work they always seem to leave QuickSync turned off so it's an "even comparison", um, what? If all I want to do is compare CPU cores alone, that's one thing, but honestly, WHY ARE YOU IGNORING these accelerators on purpose? Dr Cutress as I recall did both off and on, and points out with QSV on, it's not even close, unless you spend as much $ and considerable compute on a discrete for NVENC instead. Intel QuickAssist Technologies were a forward-thinking "bet" Intel made and released starting in 2007, by an Intel team led by an executive VP some guy named Pat Gelsinger. I wonder if anything ever came of that eh? Makes me feel like their transition to multi-faceted XPU will be another ace up their sleeve. Circling back to Patrick Kennedy's ServeTheHome article and videos, twice as many cores is nice for some workloads, meanwhile Intel performance of some very essential processes to securely hosting WWW absolutely crush it with a fraction of the cores with baked-in or optional add-on accelerators (no need to "change" hardware) and a simple driver. Dang. https://www.servethehome.com/intel-quickassist-in-ice-lake-servers-what-you-need-to-know/ https://www.servethehome.com/intel-xeon-d-2700-onboard-quickassist-qat-acceleration-deep-dive/

Add my name to the list who think LTT should build a better alternative. Heck build an incubator and support the best projects. We KNOW from how LTT have handled difficult tech issues in the past that they would totally have the best interests of the community at heart, and not put-up with any nonsense from corporate spin doctors.

For those who are interested like I am, there are some related SaaS programs out there which will machine create captions from audio (like YouTube does) and further you can edit or revise the script text and it will AI machine "learn" your voice or offer multiple "AI narrator" voices to choose from, so you can feed it the script in languages and these AI contextually figure out the context of the words. (As LTT's example show, the tone and emotion at various points can be a challenge to replicate) Couple years back part of his Masters studies this fellow created a python program that could learn your voice from a very short clip, the more samples the better it got. He partnered and went on to found Resemble.AI https://www.resemble.ai/ There was a company named themselves after the amazing Lyre bird, which can mimic humans and sounds remarkably well. It was acquired by Descript, which offers remarkable features for podcast/video auto-to-text script and then script editing are reflected in the video along with learn-your-voice overdubbing https://www.descript.com/lyrebird https://www.descript.com/ Several other "AI host from script" platforms are having success: https://www.synthesia.io/ https://elai.io/ Not trying to pick winners here, so here are some listicles: https://victorytale.com/best-ai-video-generators/ https://hourone.ai/top-7-ai-video-generators-for-2022/ https://www.makeuseof.com/best-ai-video-generators-text-to-video/

Summary Big Pharmas lose out at Cannes Lions as VMLY&R sweeps health categories with campaigns for Dell + Intel, and Maxx Flash VMLY&R awarded Pharma Grand Prix at Cannes for innovative voice-preserving tech Among 298 entries in the Pharma category this year, an innovative project designed to preserve the voices of patients who are losing them took home the Grand Prix at Cannes on Monday. UPDATE: “I Will Always be Me” won a record 8 Cannes Lions International Festival of Creativity Awards, including the granddaddy of them all, the Grand Prix. Quotes My thoughts This is fantastic example of the use of technology for good. Watch the video, I'm not crying, you're crying! Ya, ok, chokes me up. Sources https://www.thedrum.com/news/2022/06/21/vmlyr-awarded-pharma-grand-prix-cannes-lions https://www.fiercepharma.com/marketing/big-pharmas-lose-out-cannes-lions-agency-vmlyr-takes-clean-sweep-health-campaigns-dell Short version and story behind the story

It's been a short while that Youtube has introduced "Most Replayed" parts to navigation. https://www.theverge.com/2022/5/18/23117452/youtube-most-replayed-parts-of-video-feature I've searched the Forums and haven't found a thread discussing it. Is there, and the search just isn't good enough? I think maybe Linus & Luke briefly touched on it in a WAN and somewhere in a TechLinked, but, I would've thought Linus of all people would have pretty strong opinions about it one way or the other, but I don't recall seeing him do so. Maybe it's a waiting to see what the impact will be or is kind of situation. I found this discussion interesting YouTube - Pushing Pixels - Could This Feature Kill Your YouTube Channel? (PP-001)

A long time ago, in a galaxy far far away, our hero, Linus, was so blown away by the stuff he got to see on a trip to a manufacturer site, he said if they let him talk about what he wanted to, he could make four or more videos about it easy. And then... ... ... and then many many weeks later, several trips through the woods and grandmother's house and several covid waves later... ... let's slip on WAN show won't be four or more videos after all... ... and we waited ... ... and then he teased us some more. Finally we all thought, we're going to see it tomorrow. And tomorrow came. And there it was! Finally! The amazing top secret behind the scenes stuff, at least that which he could tell us about, was finally here! uhhh, wait a second. Those bunny suits cause a person's voice and accent to change? Who knew? I thought Dr Ian wasn't going to work for LTT Labs. Hmmm. With a crew of one managed to turnaround his video way faster?! Hmmmm. Must be slightly delayed yet again. Oh wait, wait, here it is, here it is! Yayyy uhhhh... Uhhh, Linus looks a lot different without the beanie hat on. And we wait.

Gizmodo's Phillip Tracy posted today regarding rumours about the next MacBook Pro 13. Summary The entry-level model will miss out on the 'Pro' features that debuted on the 14- and 16-inch models. It's an interesting enough tease of a read. In particular: Quotes My thoughts Umm, Hellloooo? Like "literally", Windows Hello. Sources https://gizmodo.com/the-next-macbook-pro-13-doesnt-sound-all-that-pro-1848491890

Summary Reports are a developer named Dmitry Selivanov has made an extension for Chrome and Firefox that can re-add the dislike count to videos across YouTube. First I noticed was feed from PCMag article Quotes My thoughts And so now the war begins between those who want the dislike count and YouTube trying to impose their will on the populace. Who will win? Sources https://www.pcmag.com/news/browser-extension-brings-back-dislike-count-to-youtube-videos

Hmmm, no it wasn't hiding. Multiple refresh was getting the same result. Then I turned off AdBlock Plus and the description and show more are back. Then I turn AdBlock Plus back on, and they're still back. It's just weird. Weirdness. So if anyone else sees such nonsense, maybe try that.

Whaaaa? It's not just me right? If you look at YouTube, and YouTubers are saying click the links in the description, and... there is a title but NO description?? Am I crazy? It's not just the Dislike count is gone, there's no extended description? Where did it go? I mean come on, that's not a small change to gloss over.

Summary Reported in PCGamer by Alex Wiltshire among others https://www.pcgamer.com/the-latency-problem-why-modern-gaming-pcs-are-slower-than-an-apple-ii/ noticing former Google and Microsoft Engineer Dan Luu couldn't help but notice old hardware seemed much more responsive that newer, he decided to test and wrote on his website. Quotes My thoughts Though counterintuitive I can almost see some engineering reasons for this to be the case. Still, nevertheless, on it's face it really is shocking and surprising. Sources http://danluu.com/input-lag/ https://www.pcgamer.com/the-latency-problem-why-modern-gaming-pcs-are-slower-than-an-apple-ii/