 |
|
|
|||||
Meta just announced a major update for its AI assistant platform, Meta AI, which has been built using the long-awaited open source Llama 3 large language model (LLM). The company says it's now the most intelligent AI assistant you can use for free. As for use case scenarios, the company touts the ability to help users study for tests, plan dinners and schedule nights out. You know the drill. Its an AI chatbot. Meta AI, however, has expanded into just about every nook and cranny throughout the companys entire portfolio, after a test run with Instagram DMs last week. Its still available with Instagram, but now users can access it on Messenger, Facebook feeds and Whatsapp. The chatbot also has a dedicated web portal at, wait for it, meta.ai. You dont need a company login to use it this way, though it wont generate images. Those recently-released Ray-Ban smart glasses also integrate with the bot, with Quest headset integration coming soon. On the topic of image generation, Meta says it's now much faster and will produce images as you type. It also handles custom animated GIFs, which is pretty cool. Hopefully, it can successfully generate images of different races of people. We found that it struggled with this basic concept a couple of weeks back, as it seemed biased toward creating images of people of the same race, even when prompted otherwise. Metas also expanding global availability along with this update, as Meta AI is coming to more than a dozen countries outside of the US. These include Australia, Canada, Ghana, Jamaica, Pakistan, Uganda and others. However, theres one major caveat. Its only in English, which doesnt seem that useful to a global audience, but whatever. As for safety and reliability, the company says Llama 3 has been trained on an expanded data set when compared to Llama 2. It also used synthetic data to create lengthy documents to train on and claims it excluded all data sources that are known to contain a high volume of personal information about private individuals. Meta says it conducted a series of evaluations to see how the chatbot would handle risk areas like conversations about weapons, cyber attacks and child exploitation, and adjusted as required. In our brief testing with the product, we've already run into hallucinations, as seen below. Engadget/Karissa Bell AI has become one of Meta CEO Mark Zuckerberg's pet projects, along with raising really expensive cattle for beef in a secluded Hawaiian compound, but the companys still playing catch up to OpenAI and, to a lesser extent, Google. Metas Llama 2 never really wowed users, due to a limited feature set, so maybe this new version of the AI assistant will catch lightning in a bottle. At the very least, it should be able to draw lightning in a bottle, or more accurately, slightly tweak someone elses drawing of lightning in a bottle.This article originally appeared on Engadget at https://www.engadget.com/meta-rolls-out-an-updated-ai-assistant-built-with-the-long-awaited-llama-3-160053435.html?src=rss
Category:
Marketing and Advertising
TSMC is the worlds biggest chipmaker and its products are found in everything from phones to game consoles and computers. But devices using TSMC chips could become more expensive if manufacturers opt to buy ones that the company makes outside of its home base of Taiwan. If a customer requests to be in a certain geographical area, the customer needs to share the incremental cost, TSMC CEO CC Wei said on an earnings call. In todays fragmented globalization environment, cost will be higher for everyone, including TSMC, our customers and our competitors. Talks with customers over price increases have already started. As the Financial Times points out, its more expensive for TSMC to manufacture chips outside of Taiwan (where over 90 percent of the planets most advanced semiconductors are made). But the company will be passing on those costs amid a push by companies and governments to increase chip supply outside of Taiwan, over which China is attempting to control. TSMC has plants in Japan and is building several in Arizona, the first of which started operating this month and is expected to go into full production this year. Its also constructing a plant in Germany. In addition, the US government last week agreed to provide the company with $6.6 billion in funding under the CHIPS Act, which seeks to bolster semiconductor manufacturing in the country. In return, TSMC pledged to up its US investment by $25 billion to $65 billion. Aligned with that, the company announced plans to build a third US plant by the end of the decade and to start making more advanced 2nm chips by 2028. Meanwhile, TSMC expects its manufacturing costs to increase in Taiwan. Thats because power prices there are soaring. An earthquake earlier this month is also expected to have a negative effect on the companys profitability, as is its struggle to make the manufacturing of its most advanced 3nm chips more efficient. Apple, NVIDIA, AMD and Qualcomm are among TSMCs more notable customers. So if they end up buying chips from the companys US, Japan or Germany fabs, their manufacturing costs could go up. Take a wild guess whod end up having to eat the cost of those increased expenses so device makers can maintain their profit margins.This article originally appeared on Engadget at https://www.engadget.com/tsmc-will-charge-more-for-chips-made-outside-of-taiwan-possibly-making-devices-more-expensive-145146879.html?src=rss
Category:
Marketing and Advertising
Theres a new invite-only app going semi-viral among VCs, tech execs and other Silicon Valley personalities. Its called Airchat and its trying to revive the concept of an audio-first social media app. The premise is similar to Clubhouse, the audio app that had a viral moment at the height of the pandemic in 2021 and inspired copycat features in Facebook, Twitter, LinkedIn and Reddit before gradually fading into obscurity. But unlike the original version of Clubhouse, Airchat isnt built around live audio streams that require users to all tune it at once. Its more like Twitter or Threads, except posts can only be shared as voice notes. The app uses a timeline format, and automatically plays audio clips as you scroll your feed. You do have the ability to pause the playback and read text instead each post is accompanied by an AI-generated transcript but posts and replies can only be shared by recording an audio clip. There dont seem to be any time constraints on how long individual clips can be, I found at least one post where a user spoke for a full hour just to see if it would work (it did). It sounds a bit gimmicky (because it is), but the app has all the hallmarks of the kind of social apps that briefly go viral among a certain segment of extremely-online Silicon Valley nerds. Its led by a pair of well-connected tech founders: AngelList co-founder Naval Ravikant and former Tinder exec Brian Norgard. Its invitation-only and has drawn a number of well-known tech personalities among its early users: Y Combinator CEO and San Francisco political provocateur Gary Tan, Spotify CEO Daniel Ek, OpenAI founder Sam Altman, VC and Elon Musk confidant Jason Calacanis. Over on X, Airchats high-profile user base is building hype and creating FOMO for those who havent been able to score an invite. (The app had to briefly shut off invites over the weekend due to an influx of new users, according to Ravikant, Airchats CEO.) Its not clear exactly how many users Airchat has just yet, but it seems to be in the low tens of thousands. The app has been downloaded close to 50,000 times, research firm Appfigures told Engadget, but its likely some of those downloads have come from people who dont yet have an invite. Ravikant, who seems to be one of the most-followed users, currently has just over 11,000 followers in the app. Using Airchat is, well, noisy and a bit confusing. Upon joining, the app asks to tap your contacts list to find friends who are already on the app, but finding people to follow beyond that can be challenging. The app doesnt have the equivalent of a for you feed with recommended content so your only options are to manually search for users or lurk in conversations that do appear in your feed and check out other users posts and follows. When I signed up, there were four people from my contacts in the app, only two of whom are actual friends. I followed them and the apps founders and a couple other familiar names. I then began randomly following other users as conversations began to appear in my feed. This was a terrible strategy as my feed was quickly dominated by the voices of a few especially active (but not terribly interesting) posters. With so many new users all joining at the same time, at one point my feed was just a bunch of people talking about Airchat. Its also somewhat jarring to actually hear the voices of people youve followed on social media for ages but havent interacted with IRL. The app defaults to playing back audio at 2x speed, which tends to make peoples speaking voices sound a bit unnatural, but is also kind of necessary for long-winded posts. The bigger issue, though, is that its not entirely clear what Airchat is for. There are a handful of channels, smaller groups dedicated to chatting about specific topics like coffee or astrology or AI or war, but conversations are disjointed and hard to follow. There seem to be some corners with spirited discussion. The coffee channel has 755 members and has lots of earnest discussion of pour-over techniques and photos of latte art. The channel is also moderated heavily, according to Ravikant (Airchats moderation policy is self moderation, which means they expect you to make good use of blocking and muting features, though an FAQ states they will remove users for harassment, impersonation, foul behavior, and illegal content.) More creative users are also finding ways to play with the audio-centric format. I found an ASMR group that consisted mainly of people speaking in breathy whispers that kind of gave me the ick (one person did post a nice clip of their cat purring). I listened to a few poetry readings in the poetry channel, but didnt have the patience, even at 2x speed. Theres also a lot of talk of in-app karaoke, though I have yet to see it actually happen. Some might see these kinds of gimmicks as the start of some new paradigm, where people use their voices to unlock new ways of interacting. But all I can think about is how Clubhouse, at its peak, had similar gimmicks: in-app game shows, open mic nights and (very NSFW) moan rooms. It was new and interesting at a time when most people were stuck at home with nothing to do, but the novelty wore off quickly. While Clubhouses initial success sparked copycat features from almost every other major social media company, many of those have since shut down due to lack of interest. Even Clubhouse itself is a shell of what it once was. While the app still exists, its an entirely different service than the one that briefly captured the attention of bored tech workers. The company laid off half its staff in 2023 and has since pivoted to audio-centric group chats.This article originally appeared on Engadget at https://www.engadget.com/airchat-is-the-latest-app-trying-to-make-social-audio-cool-again-140050450.html?src=rss
Category:
Marketing and Advertising
As learning language models (LLMs) continue to advance, so do questions about how they can benefit society in areas such as the medical field. A recent study from the University of Cambridge's School of Clinical Medicine found that OpenAI's GPT-4 performed nearly as well in an ophthalmology assessment as experts in the field, the Financial Times first reported. In the study, published in PLOS Digital Health, researchers tested the LLM, its predecessor GPT-3.5, Google's PaLM 2 and Meta's LLaMA with 87 multiple choice questions. Five expert ophthalmologists, three trainee ophthalmologists and two unspecialized junior doctors received the same mock exam. The questions came from a textbook for trialing trainees on everything from light sensitivity to lesions. The contents aren't publicly available, so the researchers believe LLMs couldn't have been trained on them previously. ChatGPT, equipped with GPT-4 or GPT-3.5, was given three chances to answer definitively or its response was marked as null. GPT-4 scored higher than the trainees and junior doctors, getting 60 of the 87 questions right. While this was significantly higher than the junior doctors' average of 37 correct answers, it just beat out the three trainees' average of 59.7. While one expert ophthalmologist only answered 56 questions accurately, the five had an average score of 66.4 right answers, beating the machine. PaLM 2 scored a 49, and GPT-3.5 scored a 42. LLaMa scored the lowest at 28, falling below the junior doctors. Notably, these trials occurred in mid-2023. While these results have potential benefits, there are also quite a few risks and concerns. Researchers noted that the study offered a limited number of questions, especially in certain categories, meaning the actual results might be varied. LLMs also have a tendency to "hallucinate" or make things up. That's one thing if its an irrelevant fact but claiming there's a cataract or cancer is another story. As is the case in many instances of LLM use, the systems also lack nuance, creating further opportunities for inaccuracy.This article originally appeared on Engadget at https://www.engadget.com/gpt-4-performed-close-to-the-level-of-expert-doctors-in-eye-assessments-131517436.html?src=rss
Category:
Marketing and Advertising
There are a frankly ridiculous number of Chromebooks on sale to choose from, but Google's Chromebook Plus initiative that launched last fall has gone a long way towards standardizing some key specs for ChromeOS devices. That in turn has made it a little easier to pick ones that'll provide a consistent quality experience without breaking the bank. Acer's latest Chromebook Plus 514, which the company just announced today, is a good example at first glance, it looks like it checks most of the boxes I'm looking for when recommending a basic Chromebook that'll work for most people. To make things confusing, Acer already sells a few Chromebook Plus 514 models; this one is the CB514-4HT and is priced at $399. Look for that SKU if you want to make sure you're getting the latest one. This laptop is powered by Intel's 13th-gen Core i3-N305 processor and pairs that with 8GB of RAM and 512GB of storage that latter spec is a bit of surprise and is a lot more storage than I'd expect to see on a $400 Chromebook. To meet the required Chromebook Plus specs, this laptop includes a 1080p webcam with a privacy shutter. While the resolution is pretty solid, not all webcams are equal so we'll have to see how this one performs in real life. Acer The display is a 14-inch, 1080p touchscreen, so it's not quite as tall as the 1,920 x 1,200 screens that I've seen on a number of other Chromebook Plus laptops. But again, at the price I'm not going to complain too much. It has a decent selection of ports, too: two USB-C and USB-A slots along with a microSD card reader. I wouldn't have minded seeing HDMI here, as the USB-C ports could quickly be taken up by power and a monitor, but I just keep reminding myself this computer is only 400 bucks. Acer says that this laptop will hit stores in early May, though the specific SKU we're talking about here should also be at Costco as early as next week. The company also says it'll have some other configurations available in the near future, though they didn't say what'll change. I wouldn't be surprised to see a model with less storage or perhaps no touchscreen, which could drive the price down even more. If so, this might be a great budget option. But even as is, you should get a pretty good laptop here for the price.This article originally appeared on Engadget at https://www.engadget.com/acers-new-399-chromebook-plus-514-hits-the-mark-for-a-solid-budget-laptop-130058747.html?src=rss
Category:
Marketing and Advertising
Sites : [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] next »