Behind the responses from genAI models are testers who evaluate those answers for accuracy, but a report released this week casts doubt on the process.According to a story published on Wednesday, contractors working on Google Gemini are now being directed to evaluate AI prompts and responses in areas in which they have no background, rather than being allowed to skip them as before.This flies in the face of the Building responsibly section of the Gemini 2.0 announcement, which said, As we develop these new technologies, we recognize the responsibility it entails, and the many questions AI agents open up for safety and security. That is why we are taking an exploratory and gradual approach to development, conducting research on multiple prototypes, iteratively implementing safety training, working with trusted testers and external experts and performing extensive risk assessments and safety and assurance evaluations.Mismatch raises questionsAccording to TechCrunch, a new internal guideline passed down from Google to contractors working on Gemini has led to concerns that Gemini could be more prone to spouting out inaccurate information on highly sensitive topics, like healthcare, to regular people.It said that the new guideline reads: You should not skip prompts that require specialized domain knowledge. Contractors are instead instructed to rate the parts they understand and add a note that they lack the necessary domain knowledge for the rest.And a blog that appeared on Artificial Intelligence+ on Thursday noted that, while contractors hired by Google to support Gemini are key players in the evaluation process one of the challenges is that [they] are often required to evaluate responses that might lie outside their own areas of expertise. For instance, while some may come from technical backgrounds, the AI can produce outputs related to literature, finance, healthcare, or even scientific research.It said, this mismatch raises questions about how effectively human oversight can serve in validating AI-generated content across diverse fields.However, Google pointed out in a later statement to TechCrunch that the raters dont only review content, they provide valuable feedback on style, format, and other factors.Hidden component of genAIWhen organizations are looking to leverage an AI model, it is important to reflect on responsible AI principles, Thomas Randall, research lead at Info-Tech Research Group said Thursday.He said that there is a hidden component to the generative AI market landscape: companies that fall under the guise of reinforcement learning from human feedback (RLHF). These companies, such as Appen, Scale AI, and Clickworker, rely on a gig economy of millions of crowd workers for data production and training the AI algorithms that we find with OpenAI, Anthropic, Google, and others. RLHF companies pose issues for fair labor practices, and are scored poorly by Fairwork.Last year, Fairwork, which defines itself as an action-research project that aims to shed light on how technological changes affect working conditions around the world, released a set of AI principles that, it said, assess the working conditions behind the development and deployment of AI systems in the context of an employment relation.There is, it stated at the time, nothing artificial about the immense amount of human labor that builds, supports, and maintains AI products and services. Many workers interact with AI systems in the workplace, and many others perform the critical data work that underpins the development of AI systems.Questions to askThe executive branch of an organization looking to leverage an AI model, said Randall, needs to ask itself an assortment of questions such as does the AI model youre using rely on or use an RLHF company? If so, was the crowd worker pool diverse enough and provided sufficient expertise? How opaque was the training process for the models you are using? Can you trace data production? If the AI vendor does not know the answers to these questions, the organization needs to be prepared to take on accountability for any outputs the AI models provide.Paul Smith-Goodson, VP and principal analyst at Moor Insights & Strategy, added that it is vitally important that Retrieval Augmented Generation (RAG) be implemented, because AI models do hallucinate and it is one way to make sure that language models are putting out the right information.He echoed Rick Villars, IDC group vice president of worldwide research, who earlier this year said, more and more the solutions around RAG and enabling people to use that more effectively are going to focus on tying into the right data that has business value, as opposed to just the raw productivity improvements.A corrosive effect on workersRyan Clarkson, managing partner at the Clarkson Law Firm, based in Malibu, California, said that the rapid growth of generative AI as a business has had corrosive effects on tech workers around the world.For example, last week, workers filed a class action lawsuit through his firm against AI data processing company Scale AI, whose services include providing the human labor to label the data used in training AI models and in shaping their responses to queries.The Scale AI lawsuit alleges poor working conditions and exploitive behavior by Scale, also saying that workers responsible for generating much of its product were mischaracterized by the company as independent contractors instead of employees.

Ready for a fresh set of puzzles? Click here for the January/February Puzzle Corner, brought to you with a special Mystery Hunt twist by guest editor Dan Katz 03.This column includes solutions to three September/October 24 problems. Find solutions to the other three problems here.

Here are solutions for the three bonus problems that appeared in the September/October 2024 Puzzle Corner column we guest edited. Solutions for S/O2, S/O4, and S/O6 are below; those for S/O1, S/O3, and S/O5 can be found here.S/O2. Frank notes that a repunit Rk is a decimal integer consisting of the digit 1 repeated k times, with k > 0. For example, R1 = 1, R2 = 11, R3 = 111, etc. Let N be any integer not divisible by 2 or 5. Prove that there is a repunit divisible by N.David Dewan came up with this solution (which Richard Lipesalso found on Wikipedia): Let N be any integer not divisible by 2 or 5. Consider repunits R1 = 1, R2 = 11, R3 = 111, , RN+1 and their residues modulo N. There are at most N different residues, so the set of N+1 residues modulo N must contain at least one repeat. Assume RA mod N and RB mod N with B>A are the same. Then (RB RA) mod N = (RB-A10A) mod N = 0. Since N and 10 are relatively prime, N divides RB-A.S/O4. Frank offers this sudoku problem:Many readers tackled it successfully. For anyone who got stuck, heres the answer key:Ten years ago I was doing what is now called AI, Richard Marks 58 writes. He noted that trying all possible iterations of a sudoku problem will tie up your Cray for a week. So I personally wrote the rules and coded this little AI program that solves any sudoko by the time you have released the Solve It button. Since I wrote the program from scratch, I guess you can say I solved S/O4.S/O6. On behalf of the MIT Chess Club, Justin Zhou 25 asked how White can play and mate in two (see below).Frank Model 63 says there are two cases to consider: when Black can castle, and when Black cannot. If Black can castle, then Blacks last move had to be Pc7-c5. In this case White can take en passant, Pb5-c6. If Black castles, then White plays Pb7#. If Black can castle, but makes some other move, then White plays Rf8#.If Black cannot castle, then White plays Ke6. Black cannot escape mate on the next move when White plays Rf8#.Steve Gordon noted that this problem illustrates the three special chess moves. The following was adapted from his analysis with his algebraic chess notation.Black last moved either a king, rook, or c5. If c5, then 1. bxc6 (en passant a.k.a e. p.). If Black can still castle queenside (1. Kc8 & Rd8, a.k.a O-O-O), then 2. b7#. If Black cannot, its king is still trapped on rank 8, so after any black move, 2. Rf8#. If Black last moved a king or rook, en passant is not possible for White, but Black cant castle either, so 1. Ke6 also traps the black king on rank 8, and after any Black move, 2. Rf8#. Note: To create this puzzle, the white bishop on g7 resulted from an underpromotion on h8.

At four years old, AirPods Max feel like a stagnant product that should be so much more, but the premium sound and Apple ecosystem features keep them from going completely stale.AirPods Max review: missing out on the latest techDecember 15, 2020 was a lifetime ago in reality and in technology. The AirPods Max were revealed shortly after the iPhone 12 lineup, when the Lightning port was still going strong and Apple Intelligence was a twinkle in Tim Cook's eye.As is usual, rumors began pouring out about a potential AirPods Max 2. The obvious features were USB-C, better chips for processing, and a folding design. Continue Reading on AppleInsider | Discuss on our Forums

A new leak claims that recent rumors of at least one iPhone 17 model switching to a horizontal row of cameras, are wrong.Close up on the iPhone's familiar camera triangleThe original rumor came from two sources that both claimed knowledge of a horizontal camera bar being introduced for at least one of the iPhone 17 models. It was suspected that the model would be the expected iPhone 17 Slim, whose thinner size is believed to be presenting engineering challenges.According to leaker "Instant Digital" on Chinese social media site Weibo, there will be a redesigned camera system but not a significant one. Rumor Score: Possible Continue Reading on AppleInsider | Discuss on our Forums

The Satechi 6-in-1 Slim USB-C adapter is small enough to fit in the pocket but provides enough connectivity to support your day-to-day needs.Satechi 6-in-1 Slim USB-C Adapter reviewAfter upgrading from an iPhone 14 to an iPhone 16 this fall, all my devices now support USB-C connections for charging and data transfer. This is also the reality for many other Apple consumers, and that's where the value of a small USB-C adapter shows itself.Satechi has been making USB-C and Thunderbolt docks and adapters for years and covers various price ranges and needs. Its latest is the 6-in-1 USB-C Slim Multiport Adapter. Continue Reading on AppleInsider | Discuss on our Forums

Marvel Rivals first seasonal event, Winter Celebration, is now live. It brings new hero outfits, including a free skin for Jeff the Land Shark. Check out the trailer below to see it in action alongside the new mode, Jeffs Winter Splash Festival.Set on Yggsgard: Royal Palace, the new mode sees two teams of Jeff players spreading snowflakes throughout the map. On top of completing missions to unlock his free skin, completing missions also earn decorations for creating a custom greeting card. As for other heroes, Groot, Rocket Raccoon, Venom, and Magik get new outfits, though they must be purchased from the in-game store. Theyre available until January 17th, 2025.Marvel Rival is available for Xbox Series X/S, PS5, and PC, having crossed 20 million players since launch. Check out our review here. Season 1 kicks off in January and will bring new maps and heroes with a Luxury Battle Pass featuring new cosmetics.

Polish studio 11 bit studios, best known forFrostpunkandThis War of Mine, has cancelled a game, codenamedProject 8, that had been in development for years.Described as a narrative-rich game that would also serve as the developers first console-centric title, 11 bit studios has announced that it has cancelled the development of Project 8. Starting development in 2018, by September 30, its development team had expanded to 37 people, while 11 bits investments into the project had grown to PLN 48.4 million (which is roughly $11.79 million).Our vision for Project 8, which was intended to be our first title designed specifically for console gamers, was bold and exciting, says 11 bit studios president Przemysaw Marsza. However, it was conceived under very different market conditions, when narrative-driven, story-rich games held stronger appeal.He went on to speak aboutthe challenges thatProject 8faced during development, from the COVID-19 pandemic to a growing budget to the aforementioned changing market conditions, all of which ultimately raised significant doubts about the projects overall profitability.WithProject 8, we experienced both breakthroughs and setbacks. During the COVID-19 pandemic, stabilising the development team proved particularly difficult, yet progress continued, he explained. While we achieved noticeable quality improvements in certain areas, several critical aspects of the game and its development process remained problematic despite multiple iterations. Over time, delays accumulated, and with each milestone, the projects budget grew.When assessing a project, the Management Board always considers its development progress, including internal reviews and independent surveys, Marsza added. We also factor in market analyses and the games sales and marketing potential. Unfortunately, in this case, our findings revealed unresolved issues and challenges that would require further extensions of the production timeline and corresponding budget increases to address. This, coupled with revised sales forecasts, largely reflecting the changing market environment, raised significant doubts about the projects overall profitability.11 bit studios says it will have to recognize an impairment loss of PLN 48.4 million due toProject 8scancellaton, while the studio is also facing layoffs as a result, with the development team working on the project being made redundant.Whether the entire teams roles are being cut or if some will be transferred to other projects within the studio hasnt been mentioned. 11 bit says it will offer severance packages and assistance in seeking new employment to those affected, and that its currently auditing the assets developed forProject 8 to assess their potential application in other initiatives.2024 saw the launch of 11 bit studiosFrostpunk 2for PC, which is also coming to PS5 and Xbox Series X/S in 2025. Next year will also see the release of the studios sci-fi survival titleThe Alters.

GSC Game World is closing the year with a major patch for S.T.A.L.K.E.R. 2: Heart of Chornobyl, but details on the future are coming soon. Along with reaffirming continued updates for the open-world shooter, it will provide a content roadmap when 2025 starts.The first round of A-Life 2.0 updates went live with yesterdays patch, so its a given that there will be more improvements. Two story expansions are also in the works, so perhaps GSC Game World has release windows and potential features to outline in its roadmap.Interestingly, the development team is also updating S.T.A.L.K.E.R.: Legends of the Zone Trilogy, the re-release of its original trilogy. It will receive a next-gen patch for consoles, while the PC version will receive more updates. Stay tuned for exact dates in the coming months.S.T.A.L.K.E.R. 2: Heart of Chornobyl is available for Xbox Series X/S, PS5, and PC. Check out our review, where we gave it a nine out of ten despite its various issues. Its sold one million copies in the first few days, with many more players joining through Game Pass.The nd of the year message from the GSC Game World team about plans for 2025. pic.twitter.com/vxzNWXIocc S.T.A.L.K.E.R. OFFICIAL (@stalker_thegame) December 20, 2024

This year saw some truly standout Unreal Engine-powered media and entertainment projects. Check out the highlights of 2024 as we look back on Dune: Part Two, Fallout, and much more.