High-quality resource bookmarks focusing on academic research | Book Marked

Google recently upgraded its AI search experience, now allowing users to directly ask follow-up questions from the “AI Overview” on the search results page and seamlessly switch to “AI Mode” for multi-turn, in-depth conversations.

(Google Logo)

At the same time, the default model for AI Overviews worldwide has been upgraded to the more powerful Gemini 3.0.

This update aims to distinguish between simple queries and complex exploratory scenarios. Users can not only quickly obtain instant information such as scores and weather but also engage in natural conversations to delve deeply into various topics.

Google stated that testing has confirmed that follow-up questions that preserve context significantly enhance the practicality of search, and the new design enables users to smoothly transition from brief summaries to deeper conversations.

This update connects with the recently launched “Personal Intelligence” feature, which leverages users’ personal data—such as Gmail and Photos—to enable the AI to provide personalized responses. These series of initiatives collectively drive Google Search’s ongoing evolution from a traditional list of results toward a dynamic, interactive intelligent assistant.

Roger Luo said:This update marks a pivotal shift of search engines from information retrieval to conversational cognitive partners. By lowering interaction barriers, Google not only improves user experience but also strengthens its strategic position as a gateway in the competitive landscape of intelligent service ecosystems.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

The plan covers 35 newly added countries and regions, having been gradually rolled out to dozens of markets since its initial launch in Indonesia last September.

The core features of the plan include access to the Gemini 3 Pro and Nano Pro models within the Gemini app, AI video creation through Veo, research and writing assistance via NotebookLM, 200GB of storage, and the ability to share benefits with up to five family members. Existing Google One Premium (2TB) users will be automatically upgraded to receive all these benefits in the coming days.

(GettyImages)

Positioned as the first upgrade option after the free tier, the plan primarily targets users who do not need or cannot afford the high-end Pro version priced at $20 per month. Its tiered pricing strategy (e.g., approximately $4.44 per month in India) directly competes with OpenAI’s ChatGPT Go plan. It aims to attract users in emerging markets and casual users with an accessible price point, fostering long-term usage habits and accelerating the adoption of AI technology and enterprise user penetration.

Roger Luo said:Google lowers the threshold for AI usage through a differentiated pricing strategy, filling the gap between the free and high-end markets with mid-range packages. This move not only directly benchmarks competitors, but also focuses on cultivating user habits in emerging markets, laying the foundation for long-term ecological layout.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

Amazon has reached a settlement in a class-action lawsuit alleging the company mishandled customer returns and refunds. The settlement is valued at over $1 billion and includes direct compensation to affected consumers as well as process improvements.

(Amazon logo)

According to court documents, the agreement stipulates:  

A cash compensation of $309.5 million to be deposited into an irrevocable common fund for the class-action suit; 

Approximately $604 million in refunds already issued or in process;  

Over $363 million in system optimization investments to enhance the returns process.

In a statement, Amazon said, “Following an internal review in 2025, we identified a small number of return orders where refunds were incomplete or where return verification was delayed. We proactively initiated refunds in that same year and will provide additional compensation in accordance with the settlement agreement.”

The lawsuit, filed in 2023, claimed that issues with Amazon’s returns system led to some consumers being charged for items they had returned, resulting in unjustified financial losses. While agreeing to the settlement, Amazon denies any wrongdoing.

Notably, this settlement marks Amazon’s second major consumer-related agreement in recent times. Last year, the company paid $2.5 billion to settle allegations by the Federal Trade Commission regarding its Prime subscription service. Claims related to that settlement are still being processed.

Sources familiar with the matter indicate that affected consumers can expect to receive notifications about compensation in the coming months. Amazon stated that it will continue to optimize its refund verification system to improve the customer experience.

Roger Luo said:This settlement highlights the systemic challenges e-commerce platforms face in large-scale operations. By proactively offering compensation and optimizing processes, Amazon demonstrates its risk management awareness. However, continuously improving operational transparency remains crucial for maintaining consumer trust and industry accountability.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

At a time when the ownership change of TikTok’s US business has caused concerns among users, the alternative application Skylight based on open source technology is experiencing rapid growth. This short video application, invested by Mark Cuba and others, and built using a decentralized AT protocol, has recently surpassed 380000 users.

(Main Photo Square)

The platform has a built-in video editor, social interaction, and community curation functions. It has accumulated over 150000 original videos and can display Bluesky content synchronously. Data shows that its daily video playback reached 1.4 million, with a growth of over 150% in new user registrations, and multiple core indicators showing multiple fold increases.

This growth wave coincides with TikTok’s completion of its US business restructuring. On January 22, TikTok announced the establishment of a new entity led by American investors, and its parent company, ByteDance, will reduce its shareholding to below 20%. The simultaneous occurrence of ownership changes and technical failures has prompted some users to switch to alternative platforms.

Roger Luo said: This trend reflects a market demand for decentralized social alternatives during ownership shifts in dominant platforms. Open-source architecture and data sovereignty are emerging as key value propositions driving user migration.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

Beijing, January 26 (Xinhua) — The international gold market once again reached a historic moment on the 26th, as the New York Mercantile Exchange gold futures price and London spot gold price simultaneously broke through the important threshold of $5100 per ounce during trading, setting a new global gold pricing record. As of the end of the trading day, the most actively traded February gold futures price reached a peak of $5108 per ounce, an increase of over 2% from the previous trading day’s closing price, demonstrating a strong and sustained upward trend.

(A gold shop assistant is organizing gold jewelry)

Market analysts generally believe that the breakthrough rise in gold prices this round is mainly driven by a combination of multiple factors. The current international geopolitical tensions continue to escalate, exacerbating uncertainty in global financial markets and prompting a large influx of safe haven funds into the precious metal market. At the same time, the adjustment of monetary policy expectations in major economies, the structural existence of global inflationary pressures, and the long-term trend of central banks continuously increasing their gold reserves have jointly built a solid support foundation for gold prices.

It is worth noting that the silver market has also shown strong performance. The price of silver futures on the New York Mercantile Exchange rose sharply during the day, breaking through the $109 per ounce mark at one point during trading and simultaneously setting a new historical high. The phenomenon of gold and silver prices rising in tandem further confirms the special position of precious metals as traditional safe haven assets in the current complex international economic environment.

Industry insiders point out that against the backdrop of fluctuations in the US dollar index, high global debt levels, and multiple countries pushing for de dollarization, the monetary attributes and asset allocation value of precious metals are being re examined by the market. In the foreseeable future, the evolution of the international political and economic situation, the policy paths of major central banks, and changes in physical demand will continue to be key variables influencing the price trends of precious metals.

Roger Luo said：Recently, international gold prices have repeatedly hit new highs, mainly driven by geopolitical tensions and market risk aversion. In this context, we can be certain that our product pricing system remains stable and has not adjusted with the drastic fluctuations in gold prices. We have always optimized supply chain management, proactively planned raw material procurement, and adhered to long-term reasonable pricing strategies, effectively buffering the cost pressure caused by external market fluctuations, and committed to providing consumers with sustainable and reliable value.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

A group of YouTube creators are suing multiple tech giants for illegally capturing their videos to train AI models, and Snap has recently been added to the list of defendants. These three plaintiffs, who collectively have approximately 6.2 million subscribers, accuse Snap of using its video content to train an AI system for in app AI features such as “Imagine Lens,” which allows users to edit images through text commands.

(evan spiegel)

Previously, the plaintiff had filed a lawsuit against Nvidia, Meta and ByteDance for similar reasons.

The latest proposed class action lawsuit was submitted to the United States District Court for the Central District of California last Friday. The plaintiff specifically pointed out that Snap used a large-scale video language dataset called HD-VILA-100M and other datasets limited to academic research purposes. The plaintiff claims that in order to use the dataset for commercial purposes, Snap circumvented YouTube’s technical restrictions, terms of service, and license provisions prohibiting commercial use.

The lawsuit demands statutory compensation and applies for a permanent injunction to prevent potential infringement in the future.

This case is mainly led by the creators of the h3h3 YouTube channel with a subscription volume of 5.52 million, as well as the smaller golf channels MrShortGame Golf and Golfholics.

This is the latest case among numerous content creators suing AI model suppliers. Previously, there have been copyright disputes from publishers, writers, newspapers, user generated content platforms, artists, and other parties. This is not the first lawsuit initiated by YouTube creators. According to data from the non-profit organization Copyright Alliance, there have been over 70 copyright infringement cases against AI companies.

The progress of such lawsuits varies: in the case of Meta and Writers Group, the judge ruled in favor of tech giants; In the case between Anthropic and the author group, the AI giant chose to settle with the plaintiff and pay compensation. Currently, the majority of cases are still under active trial.

Roger Luo said：This case centers on whether the commercial use of “research-only” datasets for AI training constitutes a substantive violation of both original content copyrights and platform terms of service. It touches on the universal legal challenge in the age of generative AI: defining the boundaries of data ownership and fair use in training materials.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

X Tests “Audio Transcription” Feature for All Video Content

(X Tests “Audio Transcription” Feature for All Video Content)

City, State – Date – X announced a new feature today. This feature transcribes audio automatically. It works for all videos uploaded to the platform. This includes videos posted by users and creators. The goal is to make video content more accessible.

The new tool creates written text from spoken words. It happens automatically after a video uploads. Users do not need to activate it manually. This text appears as captions directly on the video. People can read the captions while watching.

This feature helps many users. It benefits people who are deaf or hard of hearing. They can follow the video content through text. It also helps people watching videos in noisy places. They can mute the sound and read instead. People learning the language might find it useful too. They can hear the words and see them written down.

X developed this tool using advanced technology. It understands different accents and dialects. It also handles background noise reasonably well. The company says accuracy is a priority. They will keep improving the transcriptions over time.

Users find the feature easy to use. The captions turn on by default for new videos. Viewers can turn them off if they prefer. They can find the option in the video player settings. Content creators do not need to do anything extra. Their videos will have captions automatically.

The feature is available now. It works across the X platform. This includes the website and the mobile apps. Users see the captions on both new and existing videos. X plans to add more languages soon. This expands access for a global audience.

(X Tests “Audio Transcription” Feature for All Video Content)

This move improves accessibility significantly. It follows requests from the user community. Many users asked for better captioning tools. X listened and delivered this automated solution. The company states its commitment to inclusive features. They want everyone to enjoy the platform.

Last Saturday, a large number of Gmail users encountered abnormal email system functions, with some users experiencing chaotic email classification and abnormal spam alerts in their inbox. Google subsequently confirmed that the issue had been fully fixed.

(gmail icon)

According to the official status panel records of Google Workspace, this malfunction began around 5am Pacific Time on Saturday. Affected users have reported that a large number of emails that should have been classified under tags such as “promotion” and “social” have flooded into the main inbox, while emails from known contacts have been mistakenly marked as spam. User feedback such as’ all spam emails go straight to inbox ‘and’ filtering system suddenly crashes’ appears on social media.

During the malfunction, Google continued to update the progress of its handling, and finally announced on Saturday evening that the service had been fully restored. The official statement stated, “Some users have encountered issues with misclassification and delayed reception of emails. Emails received during the malfunction period may temporarily still display incorrect spam labels

Google stated that it will release a detailed incident analysis report after completing an internal investigation. This malfunction occurred on January 24, 2026, and all services have now resumed normal operation.

Roger Luo said:This incident exposes critical dependencies on automated filtering in large-scale systems. While swift restoration shows robust infrastructure, persistent misclassification risks eroding user trust—highlighting the need for more resilient AI-driven email management frameworks. 

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

According to Bloomberg’s renowned journalist Mark Gurman, the artificial intelligence collaboration between Apple and Google is about to achieve its first significant outcome. It is reported that Apple plans to officially release the new generation Siri in mid to late February, which will be deeply integrated with the Google Gemini artificial intelligence model.

(apple intelligence iphone mac)

It is reported that this update is expected to fulfill Apple’s technology promise made in June 2024 for the first time – the new version of Siri will be able to independently complete complex cross application operations by calling user personal data and real-time screen content. Gulman also revealed that Apple is preparing for a larger upgrade at the June Global Developers Conference: the next generation Siri will adopt a conversational interaction mode closer to ChatGPT and may be directly deployed on Google Cloud infrastructure.

Previous reports have shown that Apple has encountered bottlenecks in its self-developed artificial intelligence strategy. Gulman revealed that Apple executive Mike Rockwell even refuted his reports as “pure nonsense” during an internal meeting last summer. However, with the substantial progress of Google’s cooperation and the recent departure of John Giannandria, former head of Apple’s AI business, the tech giant’s path towards artificial intelligence development has shown a clear shift.

Roger Luo said:This collaboration reveals a new paradigm in AI competition among tech giants: Apple combines its ecosystem data with Google’s cutting-edge models, mitigating risks from in-house development delays while accelerating service intelligence. However, deep reliance on a partner’s core technology may trigger new struggles over long-term ecosystem control and data security.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.

The content of the conservative leaning AI generated encyclopedia “Grokipedia” developed by xAI, a subsidiary of Elon Musk, began to appear in ChatGPT’s responses.

(Andrey Rudakov/Bloomberg / Getty Images)

XAI launched Grokipedia in October last year, after Musk repeatedly criticized Wikipedia for bias against conservatives. The media then found that although many entries seemed to be copied directly from Wikipedia, Grokimedia also claimed that pornographic content aggravated the AIDS crisis, provided an “ideological defense” for slavery, and used derogatory expressions against cross gender groups.

For an encyclopedia derived from a chatbot that once claimed to be a “mechanical Hitler” and was used to spread deepfake pornographic content on the X platform, these contents may not be surprising. However, its information seems to be gradually spreading beyond Musk’s ecosystem – The Guardian reported that GPT-5.2 cited content from Grokipedia nine times in response to over ten different questions.

The Guardian pointed out that ChatGPT did not cite the source when asked about topics on which the false information of Grokimedia has been widely reported, such as the riots on Capitol Hill on January 6 or the AIDS epidemic. On the contrary, citations appear on more obscure topics, including statements about historian Richard Evans that The Guardian has previously clarified. Anthropic’s Claude model also referenced Grokipedia when answering certain questions. ）

A spokesperson for OpenAI told The Guardian that the company is committed to obtaining information from a wide range of publicly available sources and diverse perspectives.

Roger Luo said:This incident exposes a critical flaw in generative AI’s cross-system information integration: the absence of an effective fact-prioritization mechanism and a traceability verification framework. When algorithms indiscriminately absorb ideologically biased data sources, they not only distort the neutrality of knowledge dissemination but also risk systematically polluting the foundation of public understanding.

All articles and pictures are from the Internet. If there are any copyright issues, please contact us in time to delete.

Inquiry us

Error: Contact form not found.