Archive for the ‘Ai’ Category

Buy alert: 2 AI stocks with strong buy ratings for August 2024 – Finbold – Finance in Bold

The excitement surrounding artificial intelligence (AI) has been a major driver of the stock market rally this year, with the market value of generative AI estimated to be around $1 trillion.

AIs potential to enhance productivity and transform industries such as marketing, healthcare, and manufacturing has fueled a surge in demand for leading tech companies. This surge has significantly boosted the stock prices of firms specializing in semiconductors and servers.

In this context, Finbold analyzed the ongoing trends and tracked down the two best investments in the AI sector for the upcoming month.

As companies increasingly digitize their operations to enhance efficiency, ServiceNows (NYSE: NOW) AI and machine learning-driven workflow solutions have positioned it as a market leader.

The companys financial performance in Q2 was impressive, with subscription revenues rising by 23% on a currency-adjusted basis to $2.54 billion.

ServiceNow secured 88 high-value deals, each worth over $1 million in annual contract value, marking a 26% year-over-year increase, demonstrating strong demand among large enterprises.

Investors are particularly excited about ServiceNows integration of generative AI (GenAI), as the company aims to revolutionize workflows across industries with GenAI at its core.

Looking ahead, ServiceNows forward guidance is equally promising. The company expects Q3 sales to range between $2.66 billion and $2.665 billion, indicating a robust 20.25% annual growth at the midpoint.

Additionally, ServiceNow anticipates an operating income margin of 29.5%, reflecting strong operational efficiency. Valuation metrics also show the companys potential, with a market cap of $169.66 billion and an enterprise value of $166.50 billion.

Despite high trailing and forward PE ratios of 149.93 and 55.85 respectively, the companys PEG ratio of 2.89 suggests reasonable valuation given its growth prospects.

Analyst sentiment further supports the bullish outlook, with an average 12-month price target of $870.04, representing a 5.13% upside from the current price of $827.61.

High institutional ownership, at 90.72%, also highlights strong professional investor confidence, giving it a strong buy rating.

Micron Technology (NASDAQ: MU) has experienced a remarkable 30% stock surge this year, driven by soaring demand for its high-performance memory modules from AI servers.

As a key supplier of DRAM and NAND memory for PCs and data centers, Microns AI-tailored high-bandwidth memory chips have sold out for 2024, prompting the company to raise prices.

These price increases have significantly improved its gross margin, which rose to 27% in Q3 from 19% in Q2, with an expected rise to 34% in Q4.

This AI-fueled demand also led to an 82% year-over-year revenue increase to $6.8 billion in Q3, with projections of $7.6 billion for Q4, marking a 90% increase from the previous year.

Analysts have responded positively, with KeyCorp, Needham & Company, JPMorgan Chase (NYSE: JPM), and Barclays all raising their price targets to as high as $180 and maintaining buy or overweight ratings.

Valuation metrics include a forward PE ratio of 19.65, a PS ratio of 5.68, and a PEG ratio of 1.12, indicating robust growth potential relative to earnings.

Analyst sentiment further supports the bullish outlook, with an average 12-month price of $169.08, a high forecast of $225, and a low forecast of $140.The average price target shows a 54.54% change from the last price of $109.41.

Institutional ownership stands at 84.58%, reflecting strong confidence from professional investors, giving it a strong buy rating.

Investors looking to capitalize on the AI boom should consider these stocks, as they offer promising returns in the evolving technological landscape.

Disclaimer: The content on this site should not be considered investment advice. Investing is speculative. When investing, your capital is at risk.

Read more:

Buy alert: 2 AI stocks with strong buy ratings for August 2024 - Finbold - Finance in Bold

I just tested Asus first AMD Ryzen AI Copilot+ PC can it beat Snapdragon X Elite? Sort of – Tom’s Guide

Copilot+ PCs are about to enter their confusing era with the Asus Zenbook S 16, as we welcome x86 to the party bringing its own set of pros and cons that Ill go into.

To talk about whats going on here, the first of these new laptops packing Snapdragon X Elite rely on Arm processing. This is a mobile-first chip architecture that breaks down complex tasks into the barebones instructions and completes a single one with every tick of that processors clock cycle (otherwise known as Reduced Instruction Set Computing (RISC).

And with this choice, you get key benefits that are important to laptop users like striking that fine balance between maximizing power and giving you a nice, long battery life (as can be seen in the battery behemoth that is the HP Omnibook X).

Meanwhile, at Computex 2024, we learnt more about what Intel and AMD is bringing to the mix in terms of its x86 chipsets the architecture used for Windows PCs for over 30 years that uses Complex Instruction Set Computing (CISC) to tackle every part of a task equally. Its done the job well over those three decades, but has historically been the source behind a lot of Windows laptop battery life woes.

It made me nervous-excited in the build up to actually being able to test a laptop with the AMD Ryzen AI 9 HX 370 (what a name), and some of my nervousness has been proven right. While the Zenbook S16 is certainly improved in the battery life department when compared with other x86 laptops, it still falls behind Arm-based systems. Not only that, but we uncovered slower general performance and disk loading speeds too.

Which leaves me feeling really conflicted, because I do like the Zenbook S16 its OLED display is a feast for the eyes, keyboard and touchpad ergonomics are on point, it is very graphically capable, you wont run into any Arm-based app compatibility issues, and I absolutely adore the ceraluminum finish and premium aesthetic.

However, while it does indeed outperform the M3 MacBook Air, the nitty gritty of the results give off the vibe that this laptop is being slightly held back by the past. It makes you really think about whether it's time for Microsoft to fully turn into Arm processing for itself and other laptop makers. And I know that sort of incendiary statement brings a tonne of work for developers turning their apps and games to Arm or relying on the Prism emulation layer.

But as the Copilot+ PC world starts to get a little more confusing, this can all be boiled down to one question you need to ask yourself: do you want compatibility comfort at the expense of worry-free battery life, or sit through the transition of several Windows apps to Arm and embrace a more perfect balance between power and stamina with a Snapdragon system?

And the more I think about it, the more Im realizing that there may only be one correct answer here

This is our first step into AMDs Strix Point, so lets get to the (ahem) point. Yes, AMD Ryzen AI 9 HX 370 is a powerhouse in plenty of areas that puts the M3 MacBook Air on notice. Not only that, but Asus attention to the aesthetics and ergonomics makes this rather nice to use.

Asus struck gold with its design refresh involving these clean lines and utilitarian branding, and the Zenbook S 16 continues this trend with a gorgeously sophisticated notebook. It all starts with that ceraluminum finish (PR-speak for a ceramic/aluminum composite), which gives the lid an incredible textured finish that feels great to the touch and eliminates any fingerprints.

Its also impressively thin and light for a 16-inch laptop, though the MacBook Air does pip it in thinness and the Surface Laptop 7 edges below in weight.

Then you open it up, and youre greeted by a mouthwatering OLED screen (more on that later) alongside a fantastic keyboard/touchpad combination. In some ways, the touchpad reminds me of the Huawei MateBook X Pro giving you controls over the brightness, volume and video scrubbing along the edges of it along with a smooth multi-touch surface.

Meanwhile, the keyboard is nicely spaced out with plenty of comfortable travel on each individual key. Put simply, youll really enjoy getting stuff done on here.

I was recently an OLED convert, and the Zenbook S 16 continues my love affair with the panel technology a flash flood of accurate color for all your productivity and entertainment purposes. While LCD continues to pip it in terms of brightness, Id happily give up a little bit of that in favor of this vividity.

In terms of that accuracy, the MacBook Airs Liquid Retina panel does come close enough that you wont really tell the difference in the sRGB gamut. But its in things like watching super colorful shows or making the most of that deep contrast ratio where the S 16 really comes into its own.

So lets dabble with the AMD Ryzen AI 9 HX 370 in here. With 12 cores and 24 threads clocked at up to 5.1 GHz, it is capable of handily defeating the M3 MacBook Air and even comes close to taking on M3 Pro. When compared to Snapdragon X Elite, the picture is a little bit mixed, but were entering an era where Windows laptops are becoming more capable of taking on Apple silicon.

Where it does beat Snapdragon, however, is in three key areas:

Of course, these 3DMark results may be skewed slightly by the fact that they rely on x86 architecture, which means the Surface Laptop 7s Snapdragon X Elite will have to run it through Prism translation. However, this is a realistic representation given the job of rebuilding a lot of apps to make the most of Arm is well underway and will take a while to finish. So a compromise in performance is expected.

A complicated picture is being painted by the new AMD chips here. On one hand, they are indeed more powerful in certain areas. But its almost as if the Zenbook S 16 is being slightly hamstrung from really competing with the Arm likes of Snapdragon X Elite and Apple silicon.

If we were in a world where all Windows laptops were x86 and Microsofts Arm efforts were still a bit of a joke, I would be here telling you how the battery life has improved over the likes of Intel Core Ultra (which it has).

But were not. Were in a new era where Windows 11 systems are capable of outlasting MacBooks, so its time to alter expectations:

As you can see, the not-so-power-efficient nature of x86 means it falls behind its competition by at least a couple of hours.

Plus, in what seems to be a symptom of most Copilot+ PCs weve tested, while thermal management has improved over last generation chips, they do still get noticeably hotter than Apples notebooks.

As you saw from the performance charts up above, the margins between AMD and Qualcomm are fine, but there is a difference here in Geekbench scores, SSD transfer speeds and the way it handles transcoding video.

In practice, these wont be the biggest dips in performance in real world use. I experienced extremely little slowdown under intense multitasking pressure. But the numbers dont lie, and with both the Arm Surface Laptop 7 and x86 Zenbook S 16 coming in at near-identical prices, you are getting a slightly better price-to-performance ratio when it comes to tackling multi-core tasks and loading up big files.

Copilot+ PCs are entering their confusing era, as not every chip will give you the same experience youd expect from reading our current crop of reviews.

The Asus Zenbook S 16 is, in many ways, a good laptop. The OLED display is a spectacle encased in that beautiful ceraluminum shell with a top notch keyboard and touchpad. Not only that, but sticking to x86 gives you no issues with app compatibility while Windows developers scramble to create Arm versions.

However, you cant stop the feeling that maybe, just maybe, Strix Point plays second fiddle to Snapdragon X Elite. I mean there are some areas where it reigns supreme, such as integrated graphics and AI processing with that larger NPU.

But the Arm variant of Copilot+ PCs strikes a better balance between performance and power efficiency something that I would pick even though running into some apps that just dont work yet can be frustrating.

Original post:

I just tested Asus first AMD Ryzen AI Copilot+ PC can it beat Snapdragon X Elite? Sort of - Tom's Guide

Department of Commerce Announces New Guidance, Tools 270 Days Following President Bidens Executive Order on AI – NIST

Credit: NicoElNino/Shutterstock

The U.S. Department of Commerce announced today, on the 270-day mark since President Bidens Executive Order (EO) on the Safe, Secure and Trustworthy Development of AI, the release of new guidance and software to help improve the safety, security and trustworthiness of artificial intelligence (AI) systems.

The departments National Institute of Standards and Technology (NIST) released three final guidance documents that were first released in April for public comment, as well as a draft guidance document from the U.S. AI Safety Institute that is intended to help mitigate risks. NIST is also releasing a software package designed to measure how adversarial attacks can degrade the performance of an AI system. In addition, Commerces U.S. Patent and Trademark Office (USPTO) issued a guidance update on patent subject matter eligibility to address innovation in critical and emerging technologies, including AI.

For all its potentially transformational benefits, generative AI also brings risks that are significantly different from those we see with traditional software. These guidance documents and testing platform will inform software creators about these unique risks and help them develop ways to mitigate those risks while supporting innovation. Laurie E. Locascio, Under Secretary of Commerce for Standards and Technology and NIST Director

Read the full Department of Commerce news release.

Read the White House fact sheet on administration-wide actions on AI.

The NIST releases cover varied aspects of AI technology. Two of them appear today for the first time: One is the initial public draft of a guidance document from the U.S. AI Safety Institute, and is intended to help software developers mitigate the risks stemming from generative AI and dual-use foundation models AI systems that can be used for either beneficial or harmful purposes. The other is a testing platform designed to help AI system users and developers measure how certain types of attacks can degrade the performance of an AI system.

Of the remaining three releases, two are guidance documents designed to help manage the risks of generative AI the technology that enables many chatbots as well as text-based image and video creation tools and serve as companion resources to NISTs AI Risk Management Framework (AI RMF) and Secure Software Development Framework (SSDF). The third proposes a plan for U.S. stakeholders to work with others around the globe on AI standards. These three publications previously appeared April 29 in draft form for public comment, and NIST is now releasing their final versions.

The two releases NIST is announcing today for the first time are:

AI foundation models are powerful tools that are useful across a broad range of tasks and are sometimes called dual-use because of their potential for both benefit and harm. NISTs AI Safety Institute has released the initial public draft of its guidelines on Managing Misuse Risk for Dual-Use Foundation Models (NIST AI 800-1), which outlines voluntary best practices for how foundation model developers can protect their systems from being misused to cause deliberate harm to individuals, public safety and national security.

The draft guidance offers seven key approaches for mitigating the risks that models will be misused, along with recommendations for how to implement them and how to be transparent about their implementation. Together, these practices can help prevent models from enabling harm through activities like developing biological weapons, carrying out offensive cyber operations, and generating child sexual abuse material and nonconsensual intimate imagery.

NIST is accepting comments from the public on the draft Managing the Risk of Misuse for Dual-Use Foundation Models until Sept. 9, 2024, at 11:59 p.m. Eastern Time. Comments can be submitted to NISTAI800-1 [at] nist.gov (NISTAI800-1[at]nist[dot]gov).

One of the vulnerabilities of an AI system is the model at its core. By exposing a model to large amounts of training data, it learns to make decisions. Butif adversaries poison the training data with inaccuracies for example, by introducing data that can cause the model to misidentify stop signs as speed limit signs the model can make incorrect, potentially disastrous decisions. Testing the effects of adversarial attacks on machine learning models is one of the goals of Dioptra, a new software package aimed at helping AI developers and customers determine how well their AI software stands up to a variety of adversarial attacks.

The open-source software, available for freedownload, could help the community, including government agencies and small to medium-sized businesses, conduct evaluations to assess AI developers claims about their systems performance. This software responds to Executive Order section 4.1 (ii) (B), which requires NIST to help with model testing. Dioptra does this by allowing a user to determine what sorts of attacks would make the model perform less effectively and quantifying the performance reduction so that the user can learn how often and under what circumstances the system would fail.

Augmenting todays two initial releases are three finalized documents:

The AI RMF Generative AI Profile (NIST AI 600-1) can help organizations identify unique risks posed by generative AI and proposes actions for generative AI risk management that best aligns with their goals and priorities. The guidance is intended to be a companion resource for users of NISTsAI RMF. It centers on a list of 12 risks and just over 200 actions that developers can take to manage them.

The 12 risks include a lowered barrier to entry for cybersecurity attacks, the production of mis- and disinformation or hate speech and other harmful content, and generative AI systems confabulating or hallucinating output. After describing each risk, the document presents a matrix of actions that developers can take to mitigate it, mapped to the AI RMF.

The second finalized publication, Secure Software Development Practices for Generative AI and Dual-Use Foundation Models (NISTSpecial Publication (SP) 800-218A), is designed to be used alongside the Secure Software Development Framework (SP 800-218). While the SSDF is broadly concerned with software coding practices, the companion resource expands the SSDF in part to address a major concern with generative AI systems: They can becompromised with malicious training data that adversely affect the AI systems performance.

In addition to covering aspects of the training and use of AI systems, this guidance document identifies potential risk factors and strategies to address them. Among other recommendations, it suggests analyzing training data for signs of poisoning, bias, homogeneity and tampering.

AI systems are transforming society not only within the U.S., but around the world. A Plan for Global Engagement on AI Standards (NIST AI 100-5), todays third finalized publication, is designed to drive the worldwide development and implementation of AI-related consensus standards, cooperation and coordination, and information sharing.

The guidance is informed by priorities outlined in the NIST-developed Plan for Federal Engagement in AI Standards and Related Tools and is tied to the National Standards Strategy for Critical and Emerging Technology. This publication suggests that a broader range of multidisciplinary stakeholders from many countries participate in the standards development process.

Here is the original post:

Department of Commerce Announces New Guidance, Tools 270 Days Following President Bidens Executive Order on AI - NIST

How I Discovered Phillys Most Successful Cookbook Author Doesnt Actually Exist – Philadelphia magazine

Longform

Luisa Florence was a best-selling cookbook author on Amazon ... right up until she wasnt.

Illustration by Till Lauer

Its Tuesday afternoon, close to the start of service, and Im texting with Joe Cicala, chef of Cicala at the Divine Lorraine and someone who has been around the Italian food scene in Philly for more than a decade. I tell him I have a weird question. I tell him its totally okay if he says no that he will probably say no but that I have to ask. Because Joe is a good guy, he answers almost immediately.

Tuesday, 3:28 p.m. Jason:Im trying to track down a cook who may (or, more likely, may not) have worked at one of your restaurants. Wouldve likely been years ago. Her name is Luisa Florence. Ring any bells?

Like 5 minutes later Joe: Hey! Just asked Angela if she rang any bells and we couldnt remember anyone by that name. I also checked my contacts and emails and nothing came up. Sorry about that.

Jason: No worries at all. Im just checking everywhere I can think of. Thanks for helping out.

Joe: Anytime.

Joe isnt the first person Ive reached out to about this. He isnt the 10th. Ive been doing this for a couple days now phone calls, emails, text messages, to chefs I know, managers, PR people Ive asked to run down their client lists. Ive asked all of them the same thing: if theyve ever heard of a woman or known anyone who might have heard of a woman who matches a series of very specific criteria.

And so far, nothing. Nothing good, anyhow. Nothing solid. But thats okay. Im going to keep looking. Because there are a lot of restaurants in this city, and I cant check them all, but I can check a lot of them. And I want to be sure of one thing before I go any further.

I want to be sure that Luisa Florence isnt real.

I saw this thread on Twitter.

And I know I know thats the stupidest way ever to start a story, but truth is important here, and thats the truth. I was burning time on Twitter when I shouldve been doing something more productive, and I saw this thread that started:

This week, my wife and I are celebrating our anniversary. My parents ordered us a very practical, thoughtful gift on Amazon: a crockpot and a crockpot cookbook. Were thrilled. Theres just one minor issue: Im pretty sure the cookbook was written by an AI.

What followed was a brief dozen-tweet exploration by Matthew Kupfer, an investigative journalist for Voice of America News. Kupfer mostly writes about Russia and the Kremlin, organized crime, corruption heavy, serious stuff. Hes written for the Moscow Times and the Kyiv Post and the San Francisco Standard, so this, too, is a guy who probably had more important things to be doing. But instead, he, too, was on Twitter, writing about crockpots and cookbooks and the possibility (near-certainty, actually) that the one hed gotten as an anniversary gift The Complete Crockpot Cookbook for Beginners, 2024 edition, written by Luisa Florence was not, in fact, written by a person at all.

You probably know this story. It was the Thing of the Day on social media a while back. News outlets of varying respectability wrote hip-shot recaps and think pieces on the day-after or the day after that. And, weirdly, Kupfers tweet would even end up moving the ponderous levers of capitalism ever so slightly but all that happens later, and well get to it in due course.

Point is, it was a thing. Not a huge thing, but a thing. Kupfers thread got about 14,000 likes and 2,800 retweets and was seen by nearly 3.5 million people. And then, like 99.9 percent of everything that happens on the internet, Kupfers story just kind of faded into the cultural background radiation. Sank beneath the waves of virality, subsumed by the next days oddity or outrage. Everyone moved on.

Except me.

Luisa Florence and her cookbook allegedly / Images via Amazon

It wasnt really the idea of an AI cookbook that hooked me when I read Kupfers tweets. I got hung up on the author on the little details of Luisa Florences (alleged) life that crawled into my brain and just sat there, itching at me.

See, like most authors (probably all authors), Luisa had a bio. The story of her life, condensed down to fit on the back fold of a hardcover dust jacket or at the top of an Amazon page. There was a photo of her looking I dont know. Author-y, I guess? With her bangs and her reading glasses and her slightly out-of-focus pink scarf. But the words painted a better picture.

About the Author Luisa Florence is 60 and lives in Philadelphia. Shes the author of various recipe books, some of which are best sellers.

Her origins are Italians, she left Tuscany when she was only 12 years old because of her parents jobs. Together with her mum, Luisa has learnt to love the kitchen, and within the years she developed extraordinary skills within preserves.

In her pantry, there is never a shortage of products as she says you always have to be ready to welcome a group of friends for dinner.

Her passion took her to work in different restaurants, acquiring even more experience.

Through the years she specialised in different cooking techniques even thanks to new technologies. Air Fryer and Ninja Foodi, Crock Pot, are perfect examples of help to prepare fast dishes, but nevertheless tasty and original. In this way, Luisa can manage her time, between home and work.

Luisa also has a great artistic sense, and her charcuterie recipes look like board masterpieces. Even the ice creams in her Ninja Creami cookbook are original, delicious and creative.

Her dream is to dedicate herself completely to the kitchen.

Her books are her way to share with a large audience all her secrets that she has learnt in all these years at the stove.

Now, if you like words the way that I like words, you know theres a lot to love there. Theres so much about it thats delightfully wrong or charmingly clumsy. There are mistakes in grammar and punctuation, sentences lacking a basic understanding of how words go together, a strange British accent in mum and learnt and specialised, and a kind of weird counter-personal chewiness to the language that just feels alien. Like frantic love poetry run through a bad translation algorithm. Like how it sounds when your grandfather tries to use slang. I adore Her origins are Italians and her extraordinary skills within preserves and the phrase nevertheless tasty and original which I now say 10 times a day, like some kind of infectious verbal tic. But really, it was the very first sentence that caught my attention.

Luisa Florence is 60 and lives in Philadelphia.

Forever 60. Eternally 60. A perpetual grandmother, with her tasteful earrings and crepe-paper wrinkles, just sitting there, dreaming of the Piazzale Michelangelo and crockpot minestrone soup. I read that and thought to myself, If she was 60 when she published this book, shes probably still alive. If she was ever really alive.

And if she lived in Philadelphia?

That meant I could find her.

Me: But shes not real, though.

Also Me: But she might be.

Me: But shes not.

Also Me: But she might be.

Me: Look, everyone agrees that shes made up. EVERYONE. Theres nothing human about her. The writing is like an eighth-grader trying to talk their way into Harvard. The picture is what youd get if you typed My Best Friends Grandma into a generative AI program. You know you can buy artificially generated author photos? You can buy entire fake authors. In bulk. It happens all the time now.

Also Me: Sure, but tell me: Is Luisa Florence more or less real than some celebrity chef youre never going to meet? Someone who puts out a ghostwritten cookbook they never lay a finger on? Is she more or less real than Escoffier? You didnt know him. Hes been dead for almost a hundred years, so youre never going to know him. All youve ever seen of him is pictures and his recipes in Le Guide Culinaire. But you talk about that guy all the time.

Me: Youre an idiot. Shes not real.

Also Me: Yeah, but she might be.

When I call people and ask them about Luisa, a lot of them say they dont know her personally, but that yeah, maybe, they mightve heard the name before.

Someone says, She didnt work for me, but yeah. I know the name. I think she worked at the place that used to be where Louie Louie is now.

Penne. That place was called Penne. Luisa didnt work there.

Someone asks about Ralphs. No one at Ralphs knew her. Someone else asks about Saloon. I cant get anyone at Saloon to answer.

Vetri, someone else says. She worked at Vetri, didnt she?

No, she didnt.

One of my friends in PR asks Joey Baldino, because Joey has been around a while and has worked for everyone, but Joey says no. Never heard of her. Another friend trolls South Philly line crews for me, asking around, but gets nothing useful out of them, which might be because its South Philly and even if Luisa was literally standing right there next to the phone when the sous at Sals Red Gravy Heaven picked up, he wouldnt cop to knowing her. No one would. But it might also be that I just havent asked in the right places yet.

I get one kitchen manager on the phone, and he says this is the stupidest fucking question hes ever been asked, and Im like, Come on really? This is Philadelphia, asshole, and if THIS is the stupidest question anyone has ever asked you, then you gotta get out more. I mean, two days ago, one of my neighbors asked me if I thought a cat could survive jumping out of the second-floor window of her townhouse and I said yeah, sure, cats are amazing, and then she said, But what if it was carrying a whole chicken?, and that wasnt even the stupidest question Id heard THAT DAY.

But anyway, I keep calling, and I keep checking. I stare at the picture of her on my laptop, and its clear that the ears are lopsided, the earrings mismatched. Oh, and one of her shoulders appears to be missing, but not in a way that you notice right away. You kinda gotta look for it. And I do, because this is Luisa, right? No matter what comes of all these texts and calls and emails, this is her. Maybe the only version of her there is. Maybe just one of a billion-billion instants frozen from out of 60 years of life in Tuscany and Philly, in restaurant kitchens and her mothers kitchen and her own kitchen. Sixty years of long shifts and bad bosses, growing old in an industry that doesnt take well to age.

Its her. The author photo of her shows all the hallmarks of an AI-generated picture the shallow focus, the botched details, a kind of uncanny deadness around the eyes that I dont think theres a word for yet. This sense of something not-real pretending at life. But its Luisa.

If a machine made this, then this is the extent of her: one photo, a terrible bio, a dozen cheaply printed cookbooks churned out in rapid succession that will sit now on shelves in other peoples kitchens and be forgotten. Theyll never be beloved. No one will look back, 20 years from now, and ask where Mom got that recipe for air-fryer mozzarella sticks or crockpot chocolate peanut butter cake.

People buy cookbooks for two reasons: aspiration and utility. Aspirational cookbooks are the ones with the slick, glossy pages that make your fingertips feel slippery when you touch them. Theyve got beautiful photography, complicated recipes, words of wisdom from the chef (or the ghostwriter) about life, cooking, the view of the world from the lofty, celebrated heights they occupy. These are totems, bought by home cooks in the vain hope that some little bit of the celebrity chefs talent will be absorbed just by flipping through the pages. You buy one of these books, and maybe you try to make one recipe one half-assed attempt at Nobuyuki Matsuhisas octopus tiradito or re-creating Matty Mathesons Double Beef Patty Melt with Gruyre and Molasses Bread where you have to use two English-muffin bottoms instead of the molasses bread because who has time to make their own goddamn molasses bread when theyre still trying to figure out how to deglaze a pan with fucking maple syrup, Matty but really, you just stick it on a shelf in your kitchen and it sits there, like some kind of religious icon to be prayed to when the sauce breaks or the bchamel burns. The Saint of Lowered Expectations.

Utilitarian cookbooks are the opposite. These are the dog-eared, sauce-stained workhorses of the home kitchen. Theyre not pretty (generally), but they exist to teach you how to do a thing as quickly and simply as possible. How to make chili. How to bake a chicken. Utilitarian cookbooks are what Luisa Florence writes. Her books were (are) an attempt at teaching people how to use their crockpots, get the most out of their Ninja Foodi air fryers, can their own vegetables. And some of those books really are best-sellers on Amazon. On the day I was looking, one of Luisas slow-cooker books was number one in the Black & African American Cooking category and number six in Hungarian Cooking, Food & Wine, concurrently. And those are two audiences that dont traditionally see a lot of overlap. The shared portion of their Venn diagram is very small. So really, Luisa was bringing people together. And that would all be great, except for one thing.

Her books are really, really bad.

The 2023 edition of the Ultimate Crockpot Cookbook claims to contain 1,001 recipes, but really, there are 424.

Theres a recipe for bacon baked potatoes that, when followed, results in a kind of potato soup. Theres a recipe for Collard Green Feet Saute that, thankfully, includes no feet. There are plenty of simple, straightforward recipes that would absolutely work just fine, but there are also ones that tell you to just throw a slice of deli ham and a slice of cheddar on top of raw chicken breasts swimming in vegetable stock, turn on your crockpot, and then, six hours later, presto! Chicken cordon bleu. Like magic.

Every page in Luisas cookbook has four recipes, and theres a stretch during the chapter on pork dishes where every single list of cooking directions is just some variation on Put everything in the crockpot, turn on the crockpot, cook in the crockpot for X hours, serve warm. With the exception of a recipe for marsala pork chops that includes one additional step, this goes on for 11 pages.

Theres a whole section on drinks that you can make in a crockpot, and one of them is one cup of whiskey, one cup of ginger ale, pumpkin puree, water, maple syrup, and a cinnamon stick. For the record, thats eight shots of whiskey. Plus a half-cup of pumpkin. And I dont know what kind of Halloween-obsessed alcoholics Luisa knows, but that recipe alone could operate as a kind of reverse Turing Test no actual human being would ever include a recipe like that in a cookbook meant for other humans.

Simply titled What Is Crock Pot, Luisas introduction to crockpot cookery is my favorite part of the book. In it, she explains (several times) what a crockpot is made of, how electricity works, how a crockpot uses electricity to cook things very slowly, what a crockpot is made of (again), how glass works, how a crockpot is not a pressure cooker, and, finally, the complex socio-economic pressures felt by American women in the 1940s who, when moving into the workforce for the first time due to the industrial manpower shortages caused by overseas deployments during the second World War, were suddenly required to balance full-time careers, childcare and homemaking all at the same time. Luisa sums it up in one sentence:

At that time, women were required to prepare dinner in the morning before they left for work so that when they returned in the evening, they could successfully complete the food preparation.

And then she explains what a crockpot is made of again.

If Kindles and audiobooks have drained some of the weight from books removing them from the world as physical objects and turning them into pure data then large language models, machine learning and generative AI have made actual books as delicate as moths wings. Want to write a book? You hardly have to do anything at all. Just feed a few ideas into a computer program raised on a diet of a million other books scraped and gobbled up from the internet, and itll spit out a finished manuscript that has all the characteristics of a book without actually being a book, because no one will have actually written it. It will just happen. Therell be a bunch of words, arranged into sentences, paragraphs, chapters. It will, to the best of the artificial intelligences ability, be about what you wanted it to be about cowboys or monsters or aerobics or hedge-fund management. It will read like the best version of a cowboys and monsters hedge-fund aerobics book that the program can cobble together and be structured in a way that the AIs machine-learning program has been trained to see as successful in a certain percentage of other books within its model.

Some of the many cookbooks available on Amazon that feel uncannily similar to Florences / Images via Amazon

And sure, its a little more complicated than that, but its not a lot more complicated than that. During the course of my search for Luisa, I reached out to an AI expert for a little perspective some definition of who (or what, or why) Luisa actually is. Brian Sathianathan is the co-founder of Iterate.ai, an enterprise AI applications platform which means hes a friend of the robots, one of the people very much on the side of the Luisa Florences of the world but hes been in the game a long time. Dude worked on the first iPhone for Apple. Hes got patents. Hes seen the rise of this from the floor up.

While AI-assisted books and LLM-generated content are undoubtedly changing the publishing landscape, I firmly believe that human authors still have a marketable and profitable future, Sathianathan told me, because AI-generated content, no matter how advanced, cannot replicate the emotional depth, nuance and authenticity that human authors bring to the table.

He explained that the thing that differentiates and will always differentiate human-written books from the libraries of the machines is a connection to the human experience, something that no LLM will ever understand. By sharing their personal stories, experiences and perspectives, authors can establish a unique voice and perspective that readers cant find elsewhere.

And that sounds great on paper. Thats precisely what the new gods of artificial intelligence want us all to believe. And it might might actually be true. But in order to be an author, one must first be an authority in something. The job title is right there in the word, and thats basically the only qualification. You have to know something (about cowboys, aerobics, crockpots, whatever), and you have to know enough of something to fill a book.

So by that logic, an AI is actually the ideal author. Because who is more of an authority on something than the system that has digested an entire internets worth of information on any given topic?

I believe in the idea that change is constant and we are meant to evolve, Sathianathan explained. Instead of fearing the rise of AI, we should be excited to explore its possibilities and see how it can enhance our lives. AI is a tool, just like the wheel, the internet, or any other innovation that has helped us progress. Its not here to destroy what weve built, but to help us create something new and better.

And I would argue that sauted feet and a house full of party guests ripped on pumpkin whiskey are not the hallmarks of a new or better world, but sometimes thats just the price you pay. Theres an idiot-savant quality to artificial intelligences at the moment. They know everything but have difficulty explaining it to people. They can fill a book, easy, but they stumble over the individual words. They bumble, they digress, they live. Theyre not great at being authors yet, but theyre getting better every minute of every day. And all they do is practice.

Sathianathan insisted there is a middle ground, a hybrid model where human authors can harness the blinding powers of ravenous large-language models and trained generative programs to create things that we cant yet even imagine: As Nikola Tesla said, The present is theirs; the future, for which I really worked, is mine. And that, if Im being honest, is some of the creepiest shit Ive ever heard. If I was making a movie, that would be the final line in the opening montage, spoken by the robot messiah right before its soulless legions marched off to conquer the world.

One day, Luisa is just gone.

I was making my calls, my emails, went searching for the title of one of her air-fryer books, and I see that all of her books except one (Canning and Preserving for Beginners, published in 2020) are missing from Amazon. Her author page has been pulled down. Every place that once sold her books now has them listed as unavailable, out of stock. I look, and the only thing I can find is a physical copy of one of her crockpot books available on Ebay.

I buy it immediately.

I reach out to Amazon. Im not expecting any kind of answer (because its Amazon), but someone gets back to me in minutes, schedules a call for later that afternoon, says theyll only talk on background but might be able to get me an official statement, depending on what, exactly, I want to ask about.

I want to talk about Luisa. I want to know what happened to her, but Amazon cant tell me. Not officially. Amazon cant tell me why her books have been pulled, only that they have been. Only that Amazon is aware of the Matthew Kupfer tweets and the Luisa Florence situation and is very concerned with the proliferation of AI-generated content (read: books by robots) and very concerned about customer satisfaction. So I ask if theres some kind of official policy regarding the use of AI, and Im told that yes, there is. As of last year, in order to put a cap on the number of AI-assisted books being published via Kindle Direct Publishing Amazons digital self-publishing platform authors using Amazons service are no longer allowed to publish more than three different books in a single day.

Three books.

A day.

And thats ridiculous, right? I mean, thats not a limit; its just an acceptance of a problem you have no interest in solving. And the Amazon representative and I laugh about this for a few minutes about how awful it sounds when you actually say it out loud but later, after the call is done and Im thinking about it, the nickel finally drops. Artificially capping the number of books that any one publishing account can upload per day at three isnt a way to keep human authors competitive or limit the use of artificial intelligence and LLMs on the platform. Amazon knows that battle has already been lost. The limit is only there to keep the robots from publishing three hundred a day, which is one of those things that are so grim that you have to laugh just to keep from crying.

(Later, Sathianathan would explain to me that this sort of low-quality work, dumped by the ton onto self-publishing platforms, has a name: [Its] called Content Spam. The problem is that AI-generated content is often created using templates and formulas, which can easily be replicated and scaled. This means that it can generate a large volume of content quickly and cheaply. But there are a few reasons why content spam is unlikely to become a major issue: algorithmic fatigue, lack of creativity, and readers skepticism of its quality and authenticity.)

By the end of the next day, Ashley Vanicek, Amazons spokesperson, gets me the companys official statement. It reads like this:

We aim to provide the best possible shopping, reading, and publishing experience, and we are constantly evaluating developments that impact that experience, which includes the rapid evolution and expansion of generative AI tools.

We havecontent guidelines governing which books can be listed for sale, and we have a robust set of methods that help us proactively detect content that violates our guidelines, whether AI-generated or not. We also remove books that do not adhere to those guidelines, including content that creates a poor customer experience. When patterns of abuse warrant it, we suspend publisher accounts to prevent repeated abuse.

Our process and guidelines will keep evolving as we see changes in AI-driven publishing to make sure we provide the best possible experience for customers.

Nowhere in a books description on Amazon does it have to say whether it was written using AI. Publishers are required to give that information to Amazon and Kindle Direct Publishing, but its only used internally. The customer never sees it.

The volume limits on new works published were put in place purely to protect customers. For Amazon, its a quality-control issue, not an artistic one. No one wants to tell the robots theyre bad at what theyre doing, and Amazon cant possibly police every book on its platform. The company just wants to make sure the books arent so bad that people start to complain.

Finally, Amazon confirms that the specific content Im concerned about Luisas books is no longer available anywhere on Amazon (and even in the wild, among resellers, theyre rare). Luisa has been disappeared. Shes gone. And Im a little bit sad, actually. I miss her almost as soon as I get the news. But after all this, you have to ask yourself what gone even means under these circumstances.

What does it mean to vanish when you never really existed in the first place?

A few days after I give up my search for Luisa Florence, I find something.

On the Amazon listing page for her one remaining book the canning book from 2020, which exists now solely as a fixed point in the digital landscape, not as a thing that you can actually buy theres a publisher listed: Zoe Publishing Ltd.

Id looked for Zoe Publishing earlier. Ever the hoarder of unremarkable things, I still have a yellow Post-it note on my desk with that name scribbled on it, with a question mark and some other queries I had for Vanicek, the Amazon spokesperson.

Heres the thing, though. I didnt ask her. I forgot. And while Im sure Id googled the name (once again trusting the artificial intelligence that commands search algorithms), Im also sure I didnt find anything. I would have told you if I had.

Now, though, I punch the name in, and I get a UK.gov website that appears to be some sort of massive repository of information on every business legally operating within the United Kingdom. Its a cold, sterile, towering example of bureaucratic brutalism massive, ruthlessly organized, cross-referenced and efficient. And Zoe Publishing Ltd is listed among its files.

Zoe Publishing Ltd is or was a private limited company, incorporated on the 21st of July, 2020 (just a couple months before Luisas first books were released), then dissolved on December 19th, 2023. Its listed as primarily being concerned with the publishing and retail sale of books, both through mail order and over the internet, and it has an office address listed on Great West Road in Brentford, U.K. The company has zero employees, no outstanding debt or properties, and just one officer: Luigi Sorgia, born September 1972, in Italy.

Yeah, his origins are Italians.

Online, I cant find the company, only the shrapnel of its dissolution. A few books on day-trading, online schooling, slow-cooker keto recipes, canning and preserving. The occasional publication date (always within the brief window of Zoe Publishings existence). On Amazon, there are some publishers with similar names (Zion Zoe Publications, Zoe Shakh Press, Zoe Rosie Publication), but Luigi Sorgia is attached to none of them that I can find. He had just the one company, and now not even that.

But he is a person. An actual, physical human. And he either invented Luisa, or he knew her, once upon a time. He may be the only one who really knows who or what she is.

So maybe Im not ready to give up my search for Luisa Florence just yet.

See the original post here:

How I Discovered Phillys Most Successful Cookbook Author Doesnt Actually Exist - Philadelphia magazine

AMD launches Amuse 2.0 generative AI tool for Ryzen and Radeon, features XDNA Super Resolution – VideoCardz.com

Shortly after Intel announced its AI Playground, AMD made a similar announcement.

This is the first release of the software, which is clearly labeled as Beta and in an experimental state. It may not perform as expected, but according to AMDs blog post, it should already provide a lot of fun for users. However, the user base will be limited to Ryzen and Radeon users.

The Amuse 2.0 software, developed with TensorStack, is designed to be simple to use, without the need to download a lot of external dependencies, use command lines, or run anything other than a single one-click executable. Compared to Intels AI Playground, the Amuse software does not support running chatbots based on Large Language Models. Currently, Amuse is only for generative AI for images (for now?)

According to AMD, Amuse uses Stable Diffusion models (open-weight models). The software will support AMD Ryzen AI 300 (recently launched Strix Point laptops), AMD Ryzen 8040 (Hawk Point), and Radeon RX 7000 series. The list is rather short, and its unclear why Radeon RX 6000 and below are excluded and why Ryzen 7040 (Phoenix), featuring nearly identical specs to Hawk Point, is not included either. However, it is assumed this will change in the future.

AMUE AI Tool, Source: AMD

AMD recommends 24GB of RAM or higher for Ryzen AI 300 and 32GB of RAM for Ryzen 8040. Theres no memory requirement listed for Radeon RX 7000 GPUs.

Amuse 2.0 features

It is worth noting that the tool supports something called XDNA Super Resolution, a special mode that is supposed to upscale images by a factor of 2. Here is a full list of supported features.

Source: AMD

See the rest here:

AMD launches Amuse 2.0 generative AI tool for Ryzen and Radeon, features XDNA Super Resolution - VideoCardz.com