Apple- New iPad Mini With A17 Pro Chip; Google Turns to Nuclear Power for AI; Musk’s Optimus Robots- All Human Operated; Apple Study-LLM’s Really Can’t Reason Yet
Posted: October 15, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, Robotics, Robots, technology Leave a commentApple made a surprise announcement today and bowed a new iPad Mini. Macrumors.com reports that the new 8.3 inch Mini will rock the A17 Pro chip and will be able to run Apple Intelligence. Apple claims a 30% boost in CPU performance and 25% boost in GPU performance compared to the older model. It will support the Apple Pencil Pro, and have faster WiFi, faster USB-C data transfer, a better 12MP wide camera, and has 128 Gigs of memory for the base model….double the amount from before. You can preorder today, it delivers October 23rd. $499 for WiFi only and $649 for WiFi plus cellular. Colors include blue, purple, space gray, and starlight.
Google is going nuclear…well, to power AI anyway. According to mashable.com, Google has cut a deal with California based Kairos Power to build between 6 and 7 mini-nuclear reactors to furnish ‘clean, round-the-clock-power’ to run its AI operations running and carbon free. they expect the first one to go online by 2030. Google is not alone. As we reported here earlier, Microsoft has cut a deal to reactivate the infamous Three Mile Island nuclear plant…which melted down years ago…in Pennsylvania to run Microsoft AI server farms. The plant could power 800,000 homes…but instead will have Microsoft as its sole customer. Not to be left out…Amazon also has a nuclear energy deal!
It had been pointed out by some tech reporters who attended that the Optimus robots at Tesla’s Cybercab and ‘We Robot’ event were human operated. Now, the verge.com says one of those, Robert Scoble, noted that an engineer told him the robots used AI to walk. However, Morgan Stanley analyst Adam Jonas wrote that the robots “relied on tele-ops (human intervention)”. One so-called robot even told an attendee…or the human voicing the robot said… “Today, I am assisted by a human,” adding that it’s not fully autonomous. (The voice stumbled on the word “autonomous.”) The first law of robotics says no harm to humans. Perhaps we need an addendum to the law of robotics…you shall not impersonate a robot and claim it is an actual robot.
As discussed here fairly often, Large Language Models…or AI, as it has been branded, isn’t really intelligent at this point. Despite OpenAI and Google claiming theirs have advanced ‘reasoning’ capabilities as the next big move for AI models. Arstechnica.com reports that a new study by 6 Apple engineers indicates that the mathematical “reasoning” displayed by advanced large language models can be extremely brittle and unreliable in the face of seemingly trivial changes to common benchmark problems. The fragility highlighted in these new results helps support previous research suggesting that LLMs use of probabilistic pattern matching is missing the formal understanding of underlying concepts needed for truly reliable mathematical reasoning capabilities. “Current LLMs are not capable of genuine logical reasoning. Instead, they attempt to replicate the reasoning steps observed in their training data.” An example from another article told of an expert witness in a legal case about real estate using Microsoft’s Copilot to figure the money damages in the case. The judge called him out on it…the AI basically faked the damages with its guesstimate. That jurist admonished the attorneys and warned that future AI attempted use has to be disclosed.
I’m Clark Reid and you’re ‘Technified’ for now.
Windows Adding New AI Features to Copilot Plus PCs; Epic Sues Google Again & Samsung, Too; Apple Backs Out of Backing OpenAI; Cruise Dinged for $1.5 Million Over Hiding Pedestrian Crash Details
Posted: October 1, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, copilot, Microsoft, microsoft-365 Leave a commentMicrosoft has laid out what they are calling the ’next phase’ of Copilot, with a new design and features bowing. 9to5google.com reports that one is ‘Copilot Voice’ which is a new experience on mobile devices that is like Google’s Gemini Live. Another new feature is ‘Copilot Daily,’ which gives a summery of news and weather using the same voice as ‘Voice,’ and grabs from ‘authorized content sources,’ while ‘Personalized Discover’ helps guide users through Copilot features. ‘Copilot Vision’ will help you understand what you are looking at on your screen and let you ask questions. One might be why does Copilot think I’m so stupid, I don’t know what I’m looking at on my screen! Microsoft does say that the controversial Recall will be available starting next month….with new privacy and security measures in place.
Epic won a case that had dragged on for 4 years against Google last December. Now, they are suing Google again, and also suing Samsung. According to theverge.com, Epic accuses Google and Samsung of illegally conspiring to undermine third party app stores. This suit flows from Samsung’s ‘Auto Blocker’ feature, that now comes on by default in new Samsung phones. While it’s turned on, it automatically keeps users from installing apps unless they come from “authorized sources” — namely, Google and Samsung’s app stores. Epic claims there’s no process for any rival store to become “authorized.” Epic complains that it now takes ‘an exceptionally onerous 21-step process’ to download a third party app store on a Samsung phone…although their own website says there are only 4 steps to do so. Some observers have pointed out that the Fortnite maker hasn’t shown how it has been harmed by the Auto Blocker.
OpenAI has dropped their supposed altruistic roots, and has gone all in as a for profit company. They have been in the midst of raising some $6.5 billion more to pour into ChatGPT, and Microsoft is expected to pump another billion into the venture. Nvidia is also expected to be a major contributor to the latest round. One major player has backed out of investing in OpenAI though…Apple. Arstechnica.com says Apple still does plan to offer limited ChatGPT integration into an upcoming iOS update, but Cupertino also plans to support additional AI models like Google’s Gemini down the line…think of it as offering a choice of large language models kind of like you have a choice of web browsers. Famously secretive Apple gave no reason for walking away from investing in OpenAI.
Cruise, the self-driving subsidiary of General Motors, has to cough up $1.5 million in fines to the National Highway Traffic Safety Administration. Techcrunch.com reports that the fine is part of a consent order signed that the company agreed to with the NHTSA over a pedestrian crash last year in San Francisco. Cruise left out the little detail in their report that the poor woman was dragged some 20 feet by the robotaxi. Cruise also has to submit a ‘corrective action plan’ outlining changes they will make towards better compliance.
I’m Clark Reid and you’re ‘Technified’ for now.
Google Gemini AI-Coming to Corporate Workspace; YouTube’s Conversational AI Bows for Premium Android Users; Spotify AI Playlist Feature Rolls Out in US; Blocked X Users Can Now See Your Posts Anyway
Posted: September 24, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, gemini, Google, technology Leave a commentGoogle has started making Gemini AI what they are calling a core part of the Workspace productivity suite, and the chatbot could therefore be adapted by millions more users. Theverge.com reports that the standalone Gemini app is being included as standard on Workspace Business, Enterprise, and Frontline plans starting sometime in Q4, replacing the need to purchase a separate Gemini add-on. To bolster security against malware, phishing, and other online threats, Google is also introducing a new “Security Advisor” tool that “delivers insights directly to an IT administrator’s inbox.” Security Advisor includes a range of safe browsing and data protection features for Chrome, Gmail, and Google Drive, and will be rolled out to paying Workspace customers “over the next few weeks.”
YouTube has begun rolling out its conversational AI feature that can answer questions about a video you are already watching. According to 9to5google.com, the feature is only available right now for Premium subscribers on Android in the US. So far, no word on when it will reach other users in the US or in other countries.
Spotify has expanded its AI Playlist tool availability to a number of countries, including the US, Canada, Ireland, and New Zealand. Techcrunch.com says it was already available to Premium subscribers in the UK and Australia. The feature is still in beta on both iOS and Android, but it allows users to create personalized playlists by inputting written prompts. Maybe you would like to hear Frank Sinatra’s 28 biggest hits, or all of Taylor Swift’s sets from her Eras Tour. You can also refine playlists you crate, and customize using locations, animals (really?), movie characters, colors, and emojis. The AI powered playlist creation tool lives under the ‘Your Library’ tab.
X has made blocking less useful. Engadget.com notes that blocked users will be able to see the posts of accounts that have blocked them. This has always been possible if one wanted to try hard enough…you could do it by switching accounts…and many that you would tend to block have a number of alternate accounts. Elon musk has wanted to disable the block feature on X for some time now.
I’m Clark Reid and you’re ‘Technified’ for now.
Meta Connect Coming Up; Disney & Direct TV Finally Make Nice; US Moves to Crack Down in Temu; Microsoft Launches Copilot Pages
Posted: September 16, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, copilot, Microsoft, microsoft-365 Leave a commentMeta Connect ’24 is coming up on September 25th. The two day extravaganza will lean into AI heavily, of course, as Apple just did, and as Google and Microsoft are doing. Engadget.com reports that after dropping their expensive mixed reality headset…that was aimed at the crazy expensive Apple Vision Pro, Meta will focus on their augmented reality glasses…which are code named Orion. Unlike the Quest 3, which covers all your vision and uses cams to get you a low quality view of the world, Orion may be set for you to view the real world like through a regular pair of glasses…BUT with a layer of holographic imagery on top of the reality up ahead. The company plans to release a new pair of Ray-Ban smart glasses next year, that will have a small built-in screen along with the existing camera, speaker, and microphone. Most Meta-watchers also think we will see a stripped down version of the Quest 3 called the 3S. Meta is aiming to have this sell for $300-$400. It may at least partially replace the Quest 2, which has been priced at $299 for some time now. Naturally, AI will be woven into practically everything they show or mention…much like Apple. You can’t over-buzz that buzzword, apparently.
After 2 weeks of blackout of ESPN, ABC, Disney+ and other Disney products, the House of Mouse and DirecTV finally cut a deal. According to variety.com, the deal was announced Saturday, and just in time for the first full day of college football on ABC and ESPN, not to mention the Prime Time Emmy show, which aired last night. All the Disney streams should be up and running on DirecTV now while the companies finish hammering out the final details. The new deal apparently gives users more flexible options for viewing. Meanwhile. DirecTV has boosted prices starting on October 6th.
The feds have proposed new rules that could make it harder and more expensive for Chinese e-commerce platforms like Tee Moo…or Teh Moo, depending on what you call it…to ship goods to the US. Arstechnica.com notes that the platform has been selling cheap goods using what is called the ‘de minimus exception’ that makes shipments valued at under $800 duty-free. Platforms taking advantage of the exemption can share less information on packages and dodge taxes. President Biden warned that “over the last 10 years, the number of shipments entering the United States claiming the de minimus exemption has increased significantly, from approximately 140 million a year to over 1 billion a year.” The government would exclude the exemption for goods covered by tariffs under laws from 1974 and 1962.
Microsoft has unveiled Copilot Pages today. The feature is supposed to be a so-called ‘canvas for multiplayer AI collaboration.’ Theverge.com says Pages lets you use Microsoft’s Copilot chatbot and pull responses into a new page where they can be edited collaboratively with others. Jared Spatero, corporate VP of AI said “You and your team can work collaboratively in a page with Copilot, seeing everyone’s work in real time and iterating with Copilot like a partner, adding more content from your data, files, and the web to your Page. This is an entirely new work pattern — multiplayer, human to AI to human collaboration.” Pages is rolling out to Microsoft 365 Copilot customers today and should be available to all subscribers later this month.
I’m Clark Reid and you’re ‘Technified’ for now.’
Google-New Tack on Reigning In Explicit Deepfakes; Feds Rule Amazon Responsible for Defective 3rd Party Products on Platform; Meta Blames ‘Hallucinations’-It’s AI Claimed Trump Shooting Fake; Tesla Recall-1.8 Million Vehicles
Posted: July 31, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, deepfake, deepfakes, technology Leave a commentGoogle is updating its ranking systems, in an effort to limit deepfakes. Explicit deepfakes have been a particularly vexing problem for female celebrities. Mashable.com reports that the Google change will do this: When someone uses terms to seek out nonconsensual deepfakes of specific individuals, the ranking system will attempt to instead provide “high-quality, non-explicit content,” such as news articles, when it’s available. Google product manager Emma Higham wrote in a blog post that “With these changes, people can read about the impact deepfakes are having on society, rather than see pages with actual nonconsensual fake images.” The ranking update has already decreased exposure to explicit image results on deepfake searches by 70%. Additionally, Google is updating systems that handle requests for removing nonconsensual deepfakes from Search. The changes should make the request process easier.
The US Consumer Product Safety Commission has ruled..unanimously…that Amazon is a distributor, and that it bears responsibility for faulty products it has sold on its marketplace. According to geekwire.com, this puts Amazon on the hook for recalled products sold by third-party sellers…that amounts to over half the company’s e-commerce sales. Amazon has always claimed that they shouldn’t be held liable for defective product sold by third party merchants on amazon.com…saying that the liability falls to the seller, not the marketplace ‘facilitating’ the sale. Amazon, as you might presume, plans to appeal.
We just wrote about several professors publishing a paper about AI lying last week…or ‘bullshitting,’ as they termed it. Now Meta is blaming its AI assistant’s so-called ‘hallucinations’ for saying incorrectly that the assassination attempt on former President Trump didn’t happen. Theverge.com says that the company termed the screw up ‘unfortunate’ in a company blog post. Meta claims that Meta AI was first programmed to not respond to questions about the attempted assassination but the company removed that restriction after people started noticing. It’s not just Meta that is caught up here: Google on Tuesday also had to refute claims that its Search autocomplete feature was censoring results about the assassination attempt. Since ChatGPT burst on the scene, the tech industry has been grappling with how to limit generative AI’s propensity for falsehoods. It’s apparently still hard to overcome what large language models are inherently designed to do: make stuff up.
Tesla has recalled over 1.8 million US cars, due to a risk of software not notifying individuals of a detached hood. Engadget.com notes that if a hood isn’t latched properly, it can blow up and block the driver’s field of view. Tesla is sending out an over the air software update that is supposed to fix the issue. The recall is for 2021 to 2024 Models 3, S, and X. It also includes Model Y vehicles from 2020 to 2024.
I’m Clark Reid and you’re ‘Technified’ for now.
Samsung-Big Increase in Galaxy Ring Production; Alphabet Pours $5 Billion More into Waymo; Meta Drops Biggest & Allegedly Best Open Source AI; Ocean Battery Rocks Make Oxygen
Posted: July 24, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, News, technology, Waymo Leave a commentSamsung’s Galaxy Ring appears to be a big hit, just as it has become available for general sale, following the preorder period. 9to5google..com reports that Samsung has increased production of the Ring by 150%, adding 600,000 more units this year. Samsung will crank out over a million Rings by the end of the year. That may not seem like a lot with it comes to consumer electronics, but for comparison, the number is equal to the number of Oura Rings sold in the last six years! the Ring starts at $399, and is available at several retailers in addition to Samsung’s website.
The push for self-driving vehicles rolls on. Alphabet is getting set to pump another $5 billion into its Waymo self-driving subsidiary the next several years. According to techcrunch.com, the multi year investment was announced but Ruth Porat, the company’s Chief Financial Officer. Alphabet expects with the added cash, that Waymo will continue as the world’s leading autonomous driving tech company. Waymo is presently delivering over 50,000 paid rides a week, with their fully driverless ride-hail service in San Francisco and Phoenix…and they have now added Los Angeles and Austin. Waymo just started with paid rides in LA the first of July, and will add paid trips in Austin later this year.
Meta has just released their open sourced AI model, Llama 3.1, the largest open source model ever. Meta claims that it outperforms OpenAI’s Chat GPT-4o and Anthropic’s Claude 3.5 Sonnet as measured by several benchmarks. Theverge.com notes that those large language models are private models…not open source. Meta has dropped millions into the project, and Mark Zuckerberg says that they see open source underpinning most AI moving forward, much like how Linux has become the open source operating system that powers most phones, servers, and gadgets today. Meta is working with Microsoft, Amazon, Google, Nvidia, and Databricks to help them deploy their own versions. They think their open source AI will surpass Chat GPT as the most widely used model by the end of the year.
As mining firms and nations around the world go big on deep sea mining, an interesting discovery has indicated that we should hit the brakes a bit. Thenextweb.com reports that the huge cache of potato-sized rocks on the ocean floor hold a treasure trove of manganese, nickel, and cobalt…all crucial ingredients of lithium-ion batteries. Now, scientists from the Scottish Association of Marine Science has discovered that the rocks contain a very high electric charge….like natural rock batteries. They naturally cause seawater to split into hydrogen and oxygen in a process called seawater electrolysis. It only takes 1.5 volts to split seawater…same juice as a AA battery. Considering that the rocks produce what is called ‘dark oxygen,’ that is oxygen produced without light, the scientists say we should back off a bit on the major mining of them.
I’m Clark Reid and you’re ‘Technified’ for now.
OpenAI Offering GPT-4o Mini-Cheaper, Stripped Down Version; Meta Has Looked at Investing Billions in Eyewear Giant; Amazon Partners with Better Business Bureau Over Fake Review Brokers; Apple Says It Didn’t Use YouTube Subtitles for its AI
Posted: July 18, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, chatgpt, openai, technology Leave a commentOpenAI has rolled out a stripped-down version of the ChatGPT-4o large language model. Dubbed GPT-4o Mini, it is claimed to have better accuracy than GPT-4 on tasks, and costs substantially less than GPT-3.5 Turbo. Zdnet.com reports that OpenAI is claiming that the new AI model is “the most cost-efficient small model in the market.” It is worth noting that there aren’t any parameters defining large or small models, so this may just be puffing. At any rate, GPT-4o Mini is priced at 15 cents per million input tokens and 60 cents per million output tokens…vastly more affordable than the previous models and 60% cheaper than GPT-3.5 Turbo. The model only offers text and image support right now, with audio and video expected to be added at a later date. Its training data is current through October 2023.
Meta has looked at investing billions in eyewear giant EssilorLuxottica. According to theverge.com, the blast of cash would be in furtherance of Meta’s partnership with the owner of Ray-Ban and numerous other eyewear brands. The Ray-Ban Meta smart glasses released last year now support multimodal AI to identify what wearers are seeing. They also sold more in a few months than the previous pair did in two years, according to EssilorLuxottica’s CEO.
Amazon is teaming up with the Better Business Bureau to fight fake review brokers…starting off with a lawsuit against a firm called ReviewServiceUSA.com. Geekwire.com says the suit claims that Review Services allegedly facilitates the selling of fake positive reviews for products on Amazon listings or Better Business Bureau profile pages. Amazon said it blocked more than 250 million suspected fake reviews in 2023.
We had reported yesterday that Apple was among some firms that used its OpenELM model to train AI on YouTube Subtitles. Now, 9to5mac.com reports that Apple says that’s not true. Cupertino claims that the subtitles were only used for research purposes, and that the YouTube subtitles were not used to power Apple Intelligence. Apple has said that Apple Intelligence models were trained “on licensed data, including data selected to enhance specific features, as well as publicly available data collected by our web-crawler.” Apple also stated that it has no plans to build new versions of the OpenELM model.
I’m Clark Reid and you’re ‘Technified’ for now.
FCC Wants to Make Carriers Unlock Phones After 60 Days; SCOTUS Allows Administration to Jawbone Social Platforms; Amazon Now in $2 Trillion Club; Google Translate-Support for 110 New Languages
Posted: June 27, 2024 Filed under: Uncategorized | Tags: AI, Artificial Intelligence, business, Microsoft, technology Leave a commentIn a proposed rule, called in FCC lingo a ‘Notice of Proposed Rulemaking’, the Federal Communications Commission is looking to make phone carriers unlock phones from their service after 60 days. According to techcrunch.com, the Commission plans to further study how this will work out with current plans and phone buying trends. For years, carriers have subsidized phone prices if a customer signed up for say, a two year contract…and they made up the difference in the monthly payments for the phone plus the fees for the service. Chairwoman Jessica Rosenworcel wrote “When you buy a phone, you should have the freedom to decide when to change service to the carrier you want and not have the device you own stuck by practices that prevent you from making that choice. That is why we are proposing clear, nationwide mobile phone unlocking rules.” We should know more about the proposed rule when the full notice is published in July.
The Supreme Court yesterday tossed claims that the Biden administration coerced social media platforms into censoring users by removing COVID and election-related content. Arstechnica.com reports that the 6-3 vote found that none of the plaintiffs had standing to have brought the case in the first place. One individual was suing on behalf of his brother. Justice Barrett noted in the majority opinion that Facebook had actually been taking down false info about COVID and the election before the administration even asked them to do so.
We’ve reported lately about the $2 trillion dollar companies, Microsoft, Nvidia, and Apple, and how Nvidia has passed up Apple, and even Microsoft for the top spot for a bit. Alphabet got into this high rolling club in April. Now, geekwire.com says Amazon has joined this rarified group of the world’s most valuable companies. Amazon stock is up over 30% this year, and over 50% the past 12 months. So there you have it….the top 5 most valuable companies on earth no longer include any oil companies, banks, or industrial firms…they are all tech: Microsoft, Nvidia, Apple, Alphabet, and Amazon.
Google is adding 110 new languages to Translate. It’s the biggest expansion ever for Google Translate. Theverge.com notes that there were already 133 languages supported, so this brings the total up to 243 supported languages. A number of these languages are related to others, and Google used their PaLM 2 AI language model to help add them. One of the most requested languages was added…Cantonese. Issac Caswell of Google said “Because Cantonese often overlaps with Mandarin in writing, it is tricky to find data and train models.” Caswell also said that “about a quarter of the new languages come from Africa.”
I’m Clark Reid and you’re ’Technified’ for now.
EU Says Apple Breaches the DMA-Big Fine Looms; Amazon’s AI ‘Remarkable Alexa’-for a fee; Google Bringing Gemini Access to Teens in School; Dozens of Cybertrucks Vandalized
Posted: June 24, 2024 Filed under: Uncategorized | Tags: AI, Apple, Artificial Intelligence, News, technology Leave a commentThe European Union has announced preliminary findings that Apple is in breach of its Digital Markets Act. 9to5mac.com reports that violating the DMA’s App Store anti-steering rules could get Cupertino socked with and enormous fine…up to 10% of the company’s worldwide revenue. A final decision is expected by March 2025. The DMA rules require that an app store should inform customers of alternative purchase options, direct them to those offers and make those purchases — free of charge. Apple doesn’t do that at present. The DMA does say Apple may charge a fee for facilitating ‘the initial acquisition of a new customer’ via the App Store…it just can’t charge for each ongoing transaction. Expect Apple to fight this.
Amazon is getting set to unleash what they are calling ‘Remarkable Alexa.’ The AI powered version will hit users with a monthly fee of $5-10 a month. According to androidpolice.com, there will still be a free tier of Alexa. Amazon has been playing catch up on the AI front, as has Apple. Remarkable Alexa (sorry, sounds clunky Amazon) will feature improved home automation…although users will probably have to buy additional Alexa-branded hardware to use that. The paid tier won’t be bundled with your Prime subscription, it will be on top of that $139 per year. The company has apparently set a deadline of this August for Remarkable Alexa to be ready, so you early adapters be ready to shell out another $120 a year.
Google has announced that it is bringing its Gemini AI tech to teens using their school accounts. This is in addition to allowing them to access it through personal accounts. Techcrunch.com says that Google believe3s this will help prepare them with skills they will need in the future when generative AI is more commonplace. Google also says it will not use data from chats with students to train and improve its AI models, and has taken steps to ensure it’s bringing this technology to students responsibly. Gemini has guardrails that will prevent inappropriate responses, such as illegal or age-gated substances, from appearing in responses. It will also actively recommend teens use its double-check feature to help them develop information literacy and critical thinking skills.
This was obviously not a Musk super fan or fans. Dozens of Cybertrucks in Fort Lauderdale got a paint job of sorts last week. Someone sprayed the tailgates with ‘EFF’ Elon. Theverge.com reports the trucks were being held on a public parking lot due to a reported warranty problem with the windshield wiper. The tagging was apparently easily removed from the stainless steel trucks. One reporter said it wouldn’t surprise him if Elon started selling EFF Elon decals for a price…as he has monetized insults in the past.
I’m Clark Reid and you’re ‘Technified’ for now.
clarkreid 
Recent Comments