October 28, 2023
Issue No. 76
Welcome to this week's AtherXplorer, covering the latest in gaming, blockchain, and everything in between.
This week’s top picks include:
- AI: ArtVenture Releases ‘SpookifyMe’ AI Tool; Shutterstock’s new AI-Powered Stock Image Customization
- Gaming: Sipher Odyssey’s Latest Closed Alpha Update; ‘Alan Wake 2’ Releases After 13 Years
- Web3, Metaverse, and NFTs: dYdX Releases Source Code and Launches dYdX Chain
- Legal Landscape: The UN Forms a High-level Board to Explore AI Governance, Includes OpenAI, Google, and Microsoft
ArtVenture Releases ‘SpookifyMe’ AI Tool — Get Creative for Halloween!
ArtVenture has released a new AI tool called ‘SpookifyMe’, allowing users to easily create Halloween versions of themselves. The tool lets people upload an image of themselves and then generates Halloween costumes in a few clicks.
SpookifyMe offers a variety of popular character options such as Batman, Batgirl, Spiderman, Ironman, Barbie, Harry Potter, Elsa from Frozen, Wonder Woman, and other costumes like mermaid, nurse, or police officer. Users who are feeling creative can also enter their own costume ideas.
The AI analyzes the uploaded facial image and seamlessly superimposes the selected Halloween outfit over the user. Various poses and accessories corresponding to the costume are also added. Within minutes, users can have Halloween photos of themselves as their favorite characters without any special makeup, costumes, or Photoshop skills. Best of all, SpookifyMe's costume creations can be done completely free of charge.
Early users have praised SpookifyMe's high quality and accuracy in transforming their regular photos into convincing Halloween lookalikes. The new ArtVenture tool makes it easy to celebrate Halloween in a fun way for users around the world. Further costume options and photo customizations may be added to SpookifyMe in the future.
Shutterstock Unveils AI-Powered Stock Image Customization Feature
Shutterstock has unveiled AI-driven editing tools within its massive library of over 700 million images, offering a revolutionary approach to stock photography. Using OpenAI technology, this beta feature lets users not only craft new AI-generated content but also tweak any image in Shutterstock's collection.
CEO Paul Hennessy likened this enhancement to orchestrating a personal photoshoot, narrowing the gap between photographers and creatives who license images. The AI suite introduces six main tools: Magic Brush for seamless edits, Variations for alternative image renditions, Expand Image to broaden visuals, Smart Resize for automatic dimensions, Background Remover, and the Dall-E powered AI Image Generator. A live demonstration is slated for November 9 during the "Shutterstock Showcase: Creative AI" virtual event.
Shutterstock commits to compensating artists for any licensed, edited images. However, content altered or crafted via AI won't qualify for platform licensing, ensuring the protection of artists' intellectual property and fair remuneration. Additionally, as part of its dedication to combatting online misinformation, Shutterstock has partnered with the Content Authenticity Initiative (CAI) and plans to embed CAI standards into its AI tools, promoting content authenticity and trustworthiness.
Google’s New AI-powered Features: Image Origin Verification and an Improved Maps Experience
Google is rolling out an "About this image" tool for Search that provides context on an image's origins and metadata. It shows when images were first indexed by Google and if they were AI-generated. This aims to help verify images amid spreading disinformation online. However, additional planned access methods like using Google Lens are still in development. While the tool could be useful, many images still go viral without verification.
Google is also improving its Fact Check Explorer to help fact-checkers search for image fact-checks. It's experimenting with using generative AI to provide more information on websites in its "About this result" tool.
Google is also adding a range of new AI-powered features to Maps, including more immersive navigation, easier-to-follow driving directions, and better-organized search results. The end result is an experience that will likely remind many users of Google Search. It analyzes billions of photos to return results for queries about specific items. For EV users, Maps now shows the last time EV chargers were used to verify availability.
Amazon Brings AI Product Image Generator for Advertisers and Conversational AI for Kids
Amazon has launched an AI-driven image-generation tool for advertisers. A 2023 survey found that 75% of advertisers faced challenges in creating engaging ad content. Amazon's generative AI tool aims to alleviate this by enabling brands to create lifestyle-themed images, which can boost click-through rates by up to 40%. In the Amazon Ad Console, advertisers simply choose their product and use the "Generate" function. The AI then produces brand-themed images based on the product, which can be further refined by the advertisers.
Additionally, Amazon Echo devices now have "Explore with Alexa", designed for kids under 13. It allows interactive conversations using pre-reviewed content focused mainly on animal facts. All content is generated offline and is rigorously reviewed with some AI assistance. This feature fosters a two-way interactive learning experience with Alexa. Amazon aims to further develop this feature.
Later this year, Amazon will introduce "Let's Chat", an Alexa experience for adults, ensuring that the AI doesn't use children's responses for training and sticking to standard Alexa data privacy policies. Amazon is also launching Echo Pop Kids speakers at $49.99 in the U.S. with themes like Marvel’s Avengers and Disney Princess. These speakers include a six-month Amazon Kids+ subscription. The “Explore with Alexa” feature isn't limited to kids' devices but works on any device in kids mode or with a kids’ voice ID. Amazon plans to offer more language options soon.
Lenovo’s Hybrid Enterprise AI Strategy with Nvidia
Lenovo detailed its hybrid enterprise AI strategy at its TechWorld 2023 event. The company aims to build hybrid foundation models using public LLMs fine-tuned on enterprise data with privacy controls. It also aspires to develop personal foundation models. Lenovo emphasized the importance of connecting AI systems with existing IT like ERP and CRM.
CTO Yong Rui explained Lenovo will primarily rely on partners for public LLMs. It is developing private models through fine-tuning and incorporating enterprise knowledge via a vector database. Lenovo uses a technique called Joint Fine-Tuning for private model training. It also developed joint pruning and quantization techniques to compress models for personal device usage. A key privacy component is a data management proxy module for query handling.
Lenovo extended its longtime partnership with Nvidia for its Hybrid AI vision. Hardware integrates Nvidia GPUs optimized for Nvidia AI Enterprise software. Lenovo is developing Omniverse-optimized hardware for 2024. Services will help customers identify and implement AI solutions at scale. Digital workspace offerings like Care One use generative AI for automated support. A sustainability engine empowers green IT choices and visualizes impacts through AI-powered consultants.
Apple Job Listings Suggesting AI Plans?
Apple appears poised to infuse more generative AI into its products based on recent job listings. A role on the App Store platform mentions an AI-based developer experience using generative models. Another retail role involves building a conversational AI platform for customer engagement using voice and chat. Listings also note tasks involving long-form text generation, summarization, and question-answering.
Additional listings point to fundamental model development and applications like conversational agents. Apple is also hiring to make models work locally on devices. While Apple has previously listed generative AI roles, the current openings are more specific in their requirements.
Bloomberg reported Apple plans to invest over $1 billion annually to develop generative AI features. An internal view sees not releasing such products as a "pretty big miss." Apple aims to use LLMs to power sentence completion and other features in the next iOS for Siri, Messages, and more. Exploration of generative features in apps like Xcode could help developers. AI is also being considered for playlists, Pages, and Keynote assistance.
There have been reports of an "Apple GPT" in development, but nothing has reached consumers yet. Meanwhile, competitors like Microsoft, Google, and Meta are deploying AI across hardware and software, highlighting Apple's desire to catch up in this area through generative model integration.
In Other AI News…
Sipher Odyssey’s Latest Closed Alpha Update has Arrived!
The latest Closed Alpha Update for Sipher Odyssey was released today, introducing several new features and improvements.
Some of the new features include a new Endless mode with dungeon augments that mix with the player's augment deck. The new Endless mode can be unlocked at Account Level 7. A new Banner system was added, allowing players to acquire characters, gear, and items through unique banners. A Welcome Package was also created to reward new players.
Several UI/UX improvements were made such as revising the home screen play button and XP icons. Account levels, mission rewards, and equipment stats across tiers were rebalanced. Dungeon runs were optimized to be faster. Other minor UI tweaks and notifications were updated.
Many bugs were fixed, though some issues remain such as incorrect names displaying in lobbies, long loading times, crashing when reaching high levels in Endless Mode, and visual effects persisting after battle. Models may not load properly in dungeons. Equipment can incorrectly display after being changed. Reloading can occur mid-shooting.
The update brought new gameplay elements and modes while improving balance, performance, and polish through UI revisions and bug fixes to provide an enhanced experience.
Current Closed Alpha testers can provide feedback on the changes as more players are added to help expand testing. This will help enhance the game experience for when a wider launch occurs.
New Slim PS5 Raises Concerns About Disc Drive Authentication
Fresh information has emerged about the upcoming slim version of the PlayStation 5, particularly concerning how the optional disc drive will interact with the Digital Edition of the console. A disclaimer has surfaced, causing discussions within the gaming community. It emphasizes the need for an internet connection during the setup to pair the disc drive with the PS5 console.
The slimmed-down version of the PS5 is set to offer increased storage capacity, benefiting both the standard PS5 and the PS5 Digital Edition. The enhancement provides the option to incorporate an Ultra HD Blu-ray Disc Drive into the PS5 Digital Edition.
For those choosing the PS5 Digital Edition, the Ultra HD Blu-ray Disc Drive can be acquired separately at an additional cost. Once the current inventory of the original PS5 model is depleted, the new slim PS5 will become the exclusive offering. It's essential to note that the price of the PS5 Digital model has been adjusted upwards with the introduction of the slim version, creating a premium compared to the PS5 Digital Edition.
Included with the new PS5 model is a horizontal stand. Additionally, a separate vertical stand compatible with all PS5 models will be available for purchase.
The requirement for an internet connection during the setup of the optional disc drive has raised concerns regarding the preservation of gaming experiences. The move is viewed by some as "highly concerning and very strange." The concern is primarily focused on a potential future scenario where the maintenance of authentication servers might be discontinued. While it remains possible that Sony could address this concern with a future update, the current requirement for an internet connection stands.
‘Alan Wake 2’ Finally Releases After 13 Years — “Worth the Wait”
Alan Wake 2 takes the franchise in a darker direction thirteen years after the original. Players control both Alan Wake, fighting to escape a nightmarish version of New York, and FBI profiler Saga Anderson, investigating mysterious disappearances in Bright Falls.
Combat is more survival-focused, with scarce supplies and fewer, stronger enemies. Puzzles are like Resident Evil, using clues hidden around large maps. Levels are nonlinear, blending exploration and backtracking with combat encounters.
A key innovation is the "Mind Place," an interactive 3D space accessed between levels. For Saga, it displays evidence and quests. For Alan, it's a writers' room where he arranges plot points found in-game. Switching characters is seamless.
The dichotomy between Alan's surreal urban nightmare and Saga's pastoral small-town horror gives the game a Twin Peaks-esque twined narrative. Influences include True Detective, Twin Peaks: The Return, and In the Mouth of Madness.
Clues are pieced together via "Saga Visions" profiling sequences. Alan rewrites his world using "Angel Lamps." Both affect in-game areas. The narrative implicates metafiction and alternate dimensions in a horror style.
With 28 years of experience from hits like Quantum Break and Control, the game is considered Remedy's most ambitious to date. It pays homage to Alan Wake while boldly reinventing survival horror. Though not required, familiarity with Wake and Control enhances the narrative experience.
Alan Wake 2 is praised for its atmosphere, creative innovations, and potential to open new storytelling possibilities. If its caliber is matched only every 13 years, fans may find the wait worthwhile.
In Other Gaming News…
Web3, Metaverse, and NFTs
dYdX Releases Source Code and Launches dYdX Chain
Decentralized derivatives exchange dYdX publicly released the open source code prior to its transition to its own dYdX Chain this week. This represented the first step toward the official mainnet launch. The firm introduced the open source code as a governance proposal to seek community approval.
dYdX completed the mainnet launch with validators creating the genesis block. This established the dYdX Chain as a fully operational Cosmos-based proof-of-stake blockchain secured by dYdX token stakers.
The launch concluded dYdX's move to a layer 1 solution, removing dependencies on Ethereum. Importantly, the mainnet activation meant infrastructure would now be run entirely by third-party validators and nodes rather than controlled by the company.
With both the public code release and successful mainnet launch, dYdX demonstrated significant progress toward establishing an open and decentralized protocol governed solely by its community.
Meta Reports Loss of $3.7B in Q3 in Metaverse-focused Reality Labs Division
Meta reported a $3.7 billion loss in its Reality Labs division for the third quarter due to lower Quest 2 sales ahead of the Quest 3 launch. Expenses remained flat at $4 billion while revenue dropped to $210 million. Meta expects further losses as it continues developing augmented and virtual reality products. Over the past 21 months, Reality Labs has lost around $24 billion.
Overall results were positive for Meta with third-quarter revenue rising 23% year-over-year to $34.1 billion and net income up 164%. Daily and monthly active users also grew for Facebook. Advertising and gaming performed well, partly due to China's growth. Meta is shifting focus toward generative AI and business AI.
The company launched the second generation of Ray-Ban smartglasses incorporating AI assistance. Zuckerberg said conversational AI could be key for smart glasses. Horizon software is being tested across devices and platforms. Meta intends to eventually provide AR capabilities through smart glasses.
For 2023, Meta anticipates higher infrastructure and payroll costs while Reality Labs’ losses increase substantially owing to R&D spending. Reel monetization is progressing ahead of projections. Video now comprises over half of Instagram and Facebook engagement. The guidance sets fourth-quarter revenue between $36-40 billion but acknowledges geopolitical uncertainty could impact advertising. Meta aims to tackle a "hiring underrun" from freezes while developing the metaverse experience overall.
Polygon Deploys New POL Token Contracts on Ethereum
Polygon has deployed the new POL token contracts on Ethereum, replacing the existing MATIC token. MATIC holders will have four years to migrate their holdings to POL. This development furthers Polygon's previously announced upgrade plans. Originally created as a proof-of-stake scaling solution, Polygon now encompasses various technologies for Ethereum scaling.
This includes application-specific blockchains tailored to specific uses like gaming or stablecoins, which may have their own native tokens. It also involves zero-knowledge EVM-based networks, providing scalability similar to solutions such as zkSync and Starknet. POL aims to allow users to stake their tokens across multiple compatible chains. After depositing their holdings, users can begin earning rewards and fees from each staked network.
Polygon's scaling approach has expanded significantly from its original PoS design. It now consists of a range of customized networks in addition to general scaling solutions. POL facilitates securing all of these growing systems with a unified staking method. Users are incentivized to stake across compatible chains to start benefiting from each network immediately. Going forward, Polygon plans to transition its original PoS chain to a pure zkRollup and introduce new interoperability standards connecting its various layers.
In Other Web3 & Metaverse News…
Legal Landscape, Policies, and Regulation
The UN Forms a High-level Board to Explore AI Governance, Includes OpenAI, Google, and Microsoft
The UN has formed a new high-level advisory board to explore governance strategies around artificial intelligence. Comprising 39 members from government, academia, and industry, including executives from Google and Microsoft, its goal is to advise ongoing UN AI initiatives and develop recommendations by 2024.
The advisory board, convening for the first time, will focus on building a global scientific consensus around AI risks and challenges. It will also seek to leverage AI to further the UN's Sustainable Development Goals and strengthen international cooperation. UN Secretary-General Guterres outlined AI's potential to exponentially benefit humanity by powering crisis prediction, public services, and development programs.
While some experts emphasize AI risks around security, privacy, and misinformation, the board maintains a generally constructive view of "AI for good." It aims to bridge other efforts and positions itself positively due to the UN's development ethos. However, its mandate may expand to include more critical issues. Gary Marcus, who discussed AI dangers at a recent conference, also met with UN officials, underscoring short-term fears around industrial-scale misinformation undermining democracy.
His long-term concern centers on an inability to control increasingly advanced AI, leaving open the possibility that machines act in unintended ways. Overall, the advisory board offers a high-level approach to developing international AI governance strategies. But questions remain around how exactly it and other initiatives will coordinate, as well as enforce recommendations outside their own jurisdictions.
BlackRock Fined $2.5M by SEC for Incorrect Investment Disclosure
The SEC filed charges against BlackRock for providing inaccurate disclosures about investments in one of its funds. Between 2015 and 2019, BlackRock Multi-Sector Income Trust made sizable loans to Aviron Group, a film production and advertising company. However, in annual and semi-annual reports, BlackRock incorrectly described Aviron's business as "Diversified Financial Services" and misrepresented its interest rates.
The SEC alleged that investment advisers have a responsibility to give truthful information about fund assets. BlackRock failed to do so regarding the Aviron investment. To settle the charges, BlackRock agreed to pay a $2.5 million penalty for the incorrect disclosures. While unrelated to crypto, BlackRock has been in the spotlight for its proposed Bitcoin ETF.
On the same day as the SEC's action against BlackRock, the iShares Bitcoin ETF was briefly listed on the DTCC platform. This led some to believe regulatory approval of the ETF was imminent. However, senior ETF analyst Eric Balchunas said the DTCC listing was simply part of the standard process, not an indication of approval. The ETF listing was then removed and reappeared on the DTCC hours later, confusing the cryptocurrency community. A DTCC spokesperson later clarified the ETF had been listed since August, with no implications for regulatory greenlight.
SBF Trial Continues — Caroline Ellison Thrown Under the Bus
In his first day of testimony, Sam Bankman-Fried sought to portray FTX's collapse as due to mistakes rather than malfeasance. He pointed fingers at former colleagues like Caroline Ellison of Alameda Research, his on-off girlfriend who ran the hedge fund, saying he advised her to reduce risks but she did not comply.
Bankman-Fried's lawyer walked him through FTX's early history. Regarding a controversial feature allowing Alameda to avoid liquidation, Bankman-Fried claimed it was meant to fix a bug and blamed its implementation on former executives Gary Wang and Nishad Singh.
Core to the prosecutors' case is that Bankman-Fried used Alameda to steal FTX user funds. However, he portrayed significant borrowing between the firms and FTX's ability to access user funds as normal business activities covered in the terms of service.
Bankman-Fried also sought to provide context for deleted messages by invoking a "New York Times test" from his previous job to avoid potentially damaging communications.
Regarding his famous appearance, Bankman-Fried said clothing and car choices were due to comfort and busyness rather than Ellison's stated branding reasons. He defended FTX's large sponsorship deals.
The judge allowed testimony regarding FTX's document deletion policy but barred wider defenses involving relying on counsel. Bankman-Fried aims to shift blame while portraying his actions as well-intentioned mistakes rather than criminal conduct.
In Other Legal & Regulatory News…
Subscribe to our newsletter today and be the first to receive Sipher news, AtherXplorer weekly issues, and new AtherXperiment posts. Sign up now and get ahead of the game!