Home Blog Page 90

Meta Gets Caught Gaming AI Benchmarks

Avery Winslow
-
0
Meta Gets Caught Gaming AI Benchmarks

Meta’s Llama 4 Models Spark Controversy Over Benchmark Manipulation

Meta’s recent release of two new Llama 4 models, Scout and Maverick, has sparked controversy in the AI community. Maverick, in particular, has gained attention for its impressive performance on the LMArena benchmark, a site where humans compare outputs from different systems and vote on the best one. Maverick secured the number-two spot on the leaderboard with an ELO score of 1417, surpassing OpenAI’s GPT-4 and Gemini 2.0 Flash.

The Unusual Deployment of Maverick

However, a closer look at Meta’s documentation revealed that the version of Maverick tested on LMArena was not the same as the publicly available model. The company deployed an "experimental chat version" of Maverick, specifically optimized for conversationality, which was not disclosed to the public. This has raised concerns about the fairness and reproducibility of the benchmark.

LMArena’s Response

LMArena posted on X, stating that Meta’s interpretation of their policy did not match their expectations. They acknowledged that Meta should have made it clearer that the model was customized to optimize for human preference. As a result, LMArena is updating their leaderboard policies to prevent confusion in the future.

The Concerns of Gaming the System

The controversy has sparked concerns about gaming the system and the impact on the validity of benchmarks like LMArena. When companies can submit specially-tuned versions of their models for testing while releasing different versions to the public, benchmark rankings become less meaningful as indicators of real-world performance.

AI Researcher’s Perspective

Independent AI researcher Simon Willison commented on the situation, saying, "It’s the most widely respected general benchmark because all of the other ones suck. When Llama 4 came out, the fact that it came second in the arena, just after Gemini 2.5 Pro — that really impressed me, and I’m kicking myself for not reading the small print."

Meta’s Response

Meta’s VP of generative AI, Ahmad Al-Dahle, addressed the accusations on X, stating that they had not trained on test sets and that the variable quality was due to stabilizing implementations. However, this explanation has not alleviated concerns about the company’s actions.

The Release of Llama 4

The release of Llama 4 was not without its challenges. Meta repeatedly pushed back the launch due to the model failing to meet internal expectations, which were high after the release of DeepSeek’s open-weight model.

Conclusion

The controversy surrounding Llama 4 highlights the challenges of benchmarks in the AI development process. As AI development accelerates, benchmarks are becoming battlegrounds, and companies are eager to be seen as leaders, even if it means gaming the system.

FAQs

Q: What is the controversy surrounding Llama 4?

A: The controversy surrounds the deployment of Maverick, a model that was optimized for conversationality and not disclosed to the public, which led to concerns about the fairness and reproducibility of the benchmark.

Q: What is LMArena?

A: LMArena is a site where humans compare outputs from different systems and vote on the best one.

Q: What is the ELO score?

A: The ELO score is a measure of a model’s performance in the LMArena benchmark, with a higher score indicating better performance.

Q: Why is the release of Llama 4 significant?

A: The release of Llama 4 marks a significant event in the AI development process, with many companies and researchers closely following its performance and capabilities.

Data Filter Challenge

Jordan Waverly
-
0
Data Filter Challenge

Introduction and Motivation

The rapid development of language models (LMs) has catalyzed breakthroughs across various domains, including natural language understanding, robotics, and digital human interaction. Compared with general large LMs, which are difficult to deploy on resource-constrained edge devices, edge LMs fine-tuned for target downstream tasks have the potential to achieve both greater efficiency and higher task accuracy. However, this fine-tuning hinges on the availability of high-quality, diverse datasets.
The Data Filtering Challenge for Training Edge Language Models
seeks to unite academic researchers, industry experts, and AI enthusiasts to develop data filtering techniques that refine datasets driving the next generation of edge LMs.

The Challenge

This challenge invites participants to create data filtering techniques and submit datasets refined by these methods, aiming to
significantly enhance the achievable performance of edge LMs on downstream tasks deployed on edge devices
. With a focus on improving model accuracy and applicability across crucial domains, participants will have the
opportunity to push the frontier of edge LMs and gain recognition within the AI community
. For the fine-tuning technique, we are focusing on a method known as Low-Rank Adaptation (LoRA), which allows for the creation of efficient task-specific edge LMs from pre-trained ones using fewer resources, making it ideal for devices such as smartphones and portable robots.

Methodology

The proposed methodology involves the following steps:

  1. Data Selection
  2. Data Preprocessing
  3. Data Filtering
  4. Model Fine-Tuning

Conclusion

The Data Filtering Challenge for Training Edge Language Models aims to unite researchers, industry experts, and AI enthusiasts to develop data filtering techniques that refine datasets driving the next generation of edge LMs. By leveraging LoRA and other fine-tuning techniques, participants will have the opportunity to push the frontier of edge LMs and gain recognition within the AI community.

FAQs

Q: What is the goal of the Data Filtering Challenge?

The goal is to develop data filtering techniques that refine datasets driving the next generation of edge LMs.

Q: What is LoRA and why is it used?

LoRA is a method known as Low-Rank Adaptation, which allows for the creation of efficient task-specific edge LMs from pre-trained ones using fewer resources, making it ideal for devices such as smartphones and portable robots.

Q: What are the key steps in the proposed methodology?

The key steps are data selection, data preprocessing, data filtering, and model fine-tuning.

Go Pointers Mastery

Parker Lennox
-
0
Go Pointers Mastery

Go (Golang) Pointers: A Comprehensive Guide

What is a Pointer?

A pointer is a variable that holds the memory address of another variable.

Analogy:

Think of a variable as a house and its value as a person living inside. A pointer is like the house’s address written on a note. You don’t carry the person – you carry the address.

Basic Pointer Syntax in Go

var x int = 10
var p *int = &x
  • x is an int variable.
  • &x gives the memory address of x.
  • p is a pointer to int (*int) that stores that address.

Dereferencing:

To get the value stored at the pointer address:

fmt.Println(*p) // Output: 10

Why Use Pointers?

  • Avoid copying large values (performance gain).
  • Modify values inside functions.
  • Work with dynamic data structures (like linked lists, trees).
  • Share state between functions or goroutines.

Example: Modify Value via Pointer

func update(val *int) {
    *val = 99
}

func main() {
    num := 50
    update(&num)
    fmt.Println(num) // Output: 99
}

✅ *val = 99 changes the original num since we’re working with its memory address.

Value vs Pointer

Feature Value (Copy) Pointer (Reference)
Memory usage More (copies value) Less (copies address)
Modify original? ❌ No ✅ Yes
Performance Slower with large data Faster

The new() Keyword

ptr := new(int)
*p = 77
fmt.Println(*p) // Output: 77

Pointers with Structs

type User struct {
    Name string
}

func rename(u *User) {
    u.Name = "Alice"
}

func main() {
    user := User{Name: "Bob"}
    rename(&user)
    fmt.Println(user.Name) // Output: Alice
}

✅ Modifications inside the function persist outside because we pass a pointer.

Common Pitfalls

  1. ❌ Dereferencing a nil pointer 
var p *int
fmt.Println(*p) // panic: runtime error: invalid memory address

Passing by Value vs Pointer (Recap)

Case Use Pointer?
Modify original data ✅ Yes
Pass large structs ✅ Yes
Read-only small values ❌ No

Conclusion

Pointers are a core concept in Go that unlock performance, control, and flexibility. By mastering pointers, you gain a deeper understanding of how data flows and memory works under the hood – critical skills for any serious Go developer.

✨ Remember: With great power (pointers), comes great responsibility (memory safety)!

5 Reasons I Rely on ChatGPT Daily

Emerson Blake
-
0
5 Reasons I Rely on ChatGPT Daily

Personalizing Your Conversations with ChatGPT

The Importance of Personalization

I like my AIs to know a little about me to make the conversations more personal. That doesn’t mean I’m going to share any private or confidential information. But knowing my name, my profession, and a few other tidbits can help the AI feel more like a friendly aide and less like a cold, robotic voice.

Customizing ChatGPT

With that in mind, ChatGPT lets you add certain personal details with both free and paid accounts. For this, browse to the ChatGPT website, launch the Windows program, or fire up the mobile app for iOS or Android. At the website or in the Windows program, click your profile icon in the upper right, select Settings, and then click Personalization. In the mobile app, tap the double-lined icon in the upper left, select your profile icon, and tap Personalization.

Adding Custom Details

From there, select and turn on the option for Customize ChatGPT (Custom Instructions on Android). Add your name and your occupation. You can also add certain traits you’d like ChatGPT to adopt, such as chatty, witty, serious, friendly, encouraging, imaginative, or empathetic. Further, you can include other details the AI should know about you. When done, select Save.

Using Memory Features

Next, turn on the switch for memory. Head back to the chat screen and add details or preferences at the prompt you’d like the AI to incorporate. ChatGPT suggests the following:

  • "Remember that I like concise responses."
  • "I just got a puppy!"
  • "What do you remember about me?"
  • "Where did we leave off on my last project?"

Referencing Customizations

You can then refer to your customizations in your chats. For instance, I’ve asked ChatGPT to reference my favorite TV shows or books that I’ve written. I’ve also asked for advice on dealing with my sweet tooth, my desire to get more exercise, and other goals that I’ve shared with the AI.

Conclusion

Personalizing your conversations with ChatGPT can make interactions feel more natural and engaging. By adding your name, occupation, and other details, you can create a more enjoyable experience. Experiment with the different traits and preferences to see how they affect your conversations.

Frequently Asked Questions

Q: Do I need to pay for ChatGPT to personalize my conversations?

A: No, you can personalize your conversations with both free and paid accounts.

Q: Can I change my personal details later?

A: Yes, you can go back to the Personalization settings and make changes as needed.

Q: Can I use the memory feature with free accounts?

A: Yes, the memory feature is available with both free and paid accounts.

Google is paying staff to do nothing.

Avery Winslow
-
0
Google is paying staff to do nothing.

Retaining Top AI Talent Amid Cutthroat Competition

The Battle for Talent

In the highly competitive field of artificial intelligence (AI), retaining top talent is a significant challenge. Google’s AI division, DeepMind, has resorted to using "aggressive" noncompete agreements for some AI staff in the UK, which bar them from working for competitors for up to a year.

Noncompete Agreements: A Growing Concern

The noncompete agreements in question are reportedly used for some AI staff in the UK, which is not subject to the same laws as the US. In the US, the Federal Trade Commission (FTC) banned most noncompetes last year. However, DeepMind’s London headquarters is not bound by these regulations.

The Impact on AI Researchers

The practice of noncompetes can have a significant impact on AI researchers, who may feel left out of the quick pace of AI progress during their lengthy periods of paid leave. The VP of AI at Microsoft, Nando de Freitas, recently posted on X about how DeepMind staff are reaching out to him "in despair" over the challenge of escaping their noncompete clauses.

A Selective Approach

Google has not responded to a request for comment from TechCrunch, but told Business Insider that it uses noncompetes "selectively." This raises questions about the company’s motivations behind using these agreements and whether they are truly necessary to retain top talent.

Conclusion

Retaining top AI talent is a critical issue in the competitive AI landscape. While noncompetes may be seen as a necessary measure to protect intellectual property and prevent poaching, they can have unintended consequences for AI researchers. As the industry continues to evolve, it is essential to strike a balance between protecting companies’ interests and respecting the rights and freedom of AI researchers.

Frequently Asked Questions

Q: What is a noncompete agreement?
A: A noncompete agreement is a legal contract that prohibits an individual from working for a competitor of their current or former employer for a specified period of time.

Q: Why do companies use noncompete agreements?
A: Companies use noncompete agreements to protect their intellectual property and prevent poaching of their employees by competitors.

Q: Are noncompetes legal?
A: Noncompetes are legal in some jurisdictions, but the laws surrounding them vary widely. In the US, the FTC banned most noncompetes last year, but this does not apply to DeepMind’s London headquarters.

Q: How do noncompetes affect AI researchers?
A: Noncompetes can have a significant impact on AI researchers, who may feel left out of the quick pace of AI progress during their lengthy periods of paid leave.

Tesla and BYD Diverge, GlobalFoundries Merges

Parker Lennox
-
0
Tesla and BYD Diverge, GlobalFoundries Merges

Hello from California, this is Yifan, your #techAsia host this week.

Tariffs and Tesla’s Turbulence

Yesterday was "Liberation Day" in the US, the day President Donald Trump announced a barrage of reciprocal tariffs on China, Japan, the EU and other trade partners, allies and foes alike. Meanwhile, his 25 per cent additional levies on auto imports will partially kick in on Thursday.

Trump said the tariffs will rebalance US trade relations and reduce the deficit, but his "First Buddy" Elon Musk might be the first in line to get hurt by rising trade barriers.

Musk’s Tesla has become a political symbol as the tech billionaire spearheads the so-called Department of Government Efficiency (Doge), a Trump White House task force involved in firing tens of thousands of federal workers in a push to drastically shrink the government and improve its spending efficiency. Protests against Tesla have been held across the globe and Tesla’s stock has been in free fall this year.

In the San Francisco Bay Area — the former home of Tesla headquarters and the first place in the US to embrace EVs — I’ve seen several Teslas driving down Highway 101 with their T-shaped logo removed.

Trump’s new tariffs are expected to add further headwinds for the company, particularly in its home market. While the California gigafactory might be where most of its US vehicles are finished and shipped, its auto parts supply chain will still be subject to the additional levies. Tesla shares dropped more than 8 per cent during extended trading Wednesday following the tariff announcement.

Tesla’s Turmoil and BYD’s Boom

2025 has started quite differently for BYD and Tesla. The two carmakers have been in a neck-and-neck race for the global EV crown since 2023. While both are facing headwinds, including intensifying competition, weakening economies, slowing demand and tariffs, Elon Musk is throwing a wrench in the works for Tesla.

China’s BYD has outsold Tesla in electric vehicles for a second straight quarter as the US automaker faces a backlash over CEO Elon Musk’s political activities, Nikkei Asia’s Yifan Yu reports.

For the January-March period, Tesla delivered 336,681 vehicles worldwide, down around 13 per cent compared to the same period last year. BYD, meanwhile, delivered 416,388 battery electric vehicles (BEVs) in the first three months of 2025, up 38.74 per cent year-on-year.

The Road to Profitability

Chinese autonomous driving company WeRide has said it hopes to become profitable within five years but warned that uncertainty over international government regulation and commercial partnerships make the timing "difficult to predict", write the Financial Times’ William Langley and Gloria Li.

Tony Han, founder and chief executive of the Nasdaq-listed company, said autonomous driving required huge investment and generating returns would be a "long process". The Nvidia-backed company has reported higher losses in each of the past three years.

Big to Bigger?

US contract chipmaker GlobalFoundries and United Microelectronics Corp, Taiwan’s No 2 chipmaker, are exploring the possibility of a merger amid American efforts to mitigate risks surrounding the Taiwan Strait and fend off growing competition from China in mature chips, Nikkei Asia’s Cheng Ting-Fang writes.

The tie-up would create a bigger, US-based company with a production footprint across Asia, the US and Europe. The aim of the merger would be to create a company with the economic scale to ensure America has access to mature chips as tensions simmer between China and Taiwan and as China produces more chips on its own.

Humanoids with Chinese Characteristics

Did you see the viral video in which a group of humanoid robots dance alongside human performers on Chinese state broadcaster CCTV’s Lunar New Year Gala in January? The robots might have looked a bit funny, awkwardly waving handkerchiefs up and down, but they have become a treasured priority for Beijing, Nikkei Asia’s Cissy Zhou and Ryohtaroh Satoh write.

Similar to EVs and smartphones, China’s government policies have prioritised humanoid robots as "disruptive products", with the market expected to reach $43bn by 2035.

Conclusion

The article highlights the impact of Trump’s tariffs on Tesla, the electric vehicle company led by Elon Musk, and how it may affect the company’s profitability. Additionally, it discusses the growth of BYD, a Chinese electric vehicle manufacturer, and the challenges faced by WeRide, a Chinese autonomous driving company, in achieving profitability. The article also touches on the potential merger between GlobalFoundries and United Microelectronics Corp, and the growth of humanoid robots in China.

FAQs

Q: What is the impact of Trump’s tariffs on Tesla?
A: The tariffs are expected to add further headwinds for Tesla, particularly in its home market, as its auto parts supply chain will still be subject to the additional levies.

Q: Which company has outsold Tesla in electric vehicles?
A: BYD has outsold Tesla in electric vehicles for a second straight quarter.

Q: When does WeRide expect to become profitable?
A: WeRide hopes to become profitable within five years, but warned that uncertainty over international government regulation and commercial partnerships make the timing "difficult to predict".

Q: What is the potential merger between GlobalFoundries and United Microelectronics Corp?
A: The merger would create a bigger, US-based company with a production footprint across Asia, the US and Europe, and would aim to ensure America has access to mature chips as tensions simmer between China and Taiwan and as China produces more chips on its own.

Gemini Live Rolls Out Screensharing Feature to Pixel 9 and Galaxy S25 Devices

Avery Winslow
-
0
Gemini Live Rolls Out Screensharing Feature to Pixel 9 and Galaxy S25 Devices

Google’s Gemini Live: A Revolutionary New Way to Interact with AI

Introducing Gemini Live

Google’s Gemini Live camera and screenshare functions are now rolling out on Pixel 9 series phones and Samsung Galaxy S25 devices, allowing users to ask conversational AI chatbot questions about the stuff they’re looking at in real-time. This innovative feature is part of "Project Astra," first demonstrated at Google’s I/O developer conference in May. The update is also coming soon to other Android devices, but users will need to be paid Gemini Advanced subscribers to access it.

How It Works

Once the update is available, users can activate the live video function by pushing a button and asking Gemini Live questions about whatever their camera can see. For example, they can point their camera at an aquarium tank and ask questions about specific fish. Alternatively, they can tap the new screenshare button, show Gemini Live a shopping website, and ask the AI assistant to compare products or provide styling advice.

Availability and Requirements

Gemini Live is available in 45 languages in select countries for users 18 years of age and older (excluding education and enterprise accounts). The feature started rolling out to customers last month, and some users on Reddit have confirmed it appears on their devices, including a Xiaomi phone.

Key Features

  • Point your camera at an object and ask Gemini Live questions about it
  • Tap the screenshare button and show Gemini Live a website or app to ask questions or receive assistance
  • Available in 45 languages in select countries
  • Requires a paid Gemini Advanced subscription on non-Pixel and Samsung devices

Conclusion

Gemini Live is a groundbreaking feature that revolutionizes the way we interact with AI. With its ability to answer questions in real-time about the stuff we’re looking at, it has the potential to change the way we use our devices and access information. Whether you’re looking for product recommendations or just want to learn more about the world around you, Gemini Live is a game-changer.

Frequently Asked Questions

Q: What devices support Gemini Live?
A: The feature is initially available on Pixel 9 series phones and Samsung Galaxy S25 devices, with a wider rollout to other Android devices coming soon.

Q: Do I need a subscription to use Gemini Live?
A: Yes, users on non-Pixel and Samsung devices need to be paid Gemini Advanced subscribers to access the feature.

Q: Is Gemini Live available in my country?
A: Check with Google to see if Gemini Live is available in your country. It is currently available in select countries for users 18 years of age and older.

Q: Can I use Gemini Live on my education or enterprise account?
A: No, Gemini Live is not available on education and enterprise accounts.

Meta’s Surprise Llama 4 Drop Exposes AI Ambition-Reality Gap

Emerson Blake
-
0
Meta’s Surprise Llama 4 Drop Exposes AI Ambition-Reality Gap

Overcoming the Limitations of Huge AI Models: Llama 4’s Mixture-of-Experts Architecture

Introducing Mixture-of-Experts (MoE) Architecture

Meta constructed the Llama 4 models using a mixture-of-experts (MoE) architecture, which is a way to overcome the limitations of running huge AI models. This approach is inspired by having a large team of specialized workers, where only the relevant specialists activate for a specific job. In the context of Llama 4, this means that instead of a single large model working on every task, multiple smaller models (experts) work together to achieve the same goal.

Reducing Computation Needs

The MoE architecture allows for a reduction in the computation needed to run the model, since smaller portions of neural network weights are active simultaneously. For example, Llama 4 Maverick features a 400 billion parameter size, but only 17 billion of those parameters are active at once across one of 128 experts. Similarly, Scout features 109 billion total parameters, but only 17 billion are active at once across one of 16 experts.

Current Limitations of AI Models

Current AI models have a relatively limited short-term memory. In AI, a context window acts somewhat in this fashion, determining how much information it can process simultaneously. AI language models like Llama typically process this memory as chunks of data called tokens, which can be whole words or fragments of longer words. Large context windows allow AI models to process longer documents, larger code bases, and longer conversations.

Reality Check for Llama’s Reality

Despite Meta’s promotion of Llama 4 Scout’s 10 million token context window, developers have found that using even a fraction of that amount has proven challenging due to memory limitations. Willison reported on his blog that third-party services providing access, like Groq and Fireworks, limited Scout’s context to just 128,000 tokens. Another provider, Together AI, offered 328,000 tokens.

The Need for Resources

Evidence suggests that accessing larger contexts requires immense resources. Willison pointed to Meta’s own example notebook ("build_with_llama_4"), which states that running a 1.4 million token context needs eight high-end Nvidia H100 GPUs.

Testing Troubles

Willison documented his own testing troubles. When he asked Llama 4 Scout via the OpenRouter service to summarize a long online discussion (around 20,000 tokens), the result wasn’t useful. He described the output as "complete junk output," which devolved into repetitive loops.

Conclusion

The MoE architecture used in Llama 4 models is a promising approach to overcoming the limitations of huge AI models. However, it is clear that there are still significant challenges to overcome, particularly when it comes to memory limitations and the need for resources to access larger contexts.

Frequently Asked Questions

Q: What is the Mixture-of-Experts (MoE) architecture?
A: MoE is an approach to building AI models that involves having a large team of specialized workers, where only the relevant specialists activate for a specific job.

Q: How does MoE reduce computation needs?
A: MoE reduces computation needs by having smaller portions of neural network weights active simultaneously, rather than a single large model working on every task.

Q: What is a context window in AI?
A: A context window is a measure of how much information an AI model can process simultaneously.

Q: Why is accessing larger contexts challenging?
A: Accessing larger contexts requires immense resources, including high-end GPUs and significant computational power.

Q: What are some of the limitations of Llama 4 models?
A: Llama 4 models have a relatively limited short-term memory and are challenging to use with large context windows due to memory limitations.

Meta exec denies artificially boosting Llama 4’s benchmark scores

Avery Winslow
-
0
Meta exec denies artificially boosting Llama 4’s benchmark scores

Rumor Denied: Meta Executive Addresses Concerns Over AI Model Benchmarking

Meta Denies Training AI Models on Test Sets

A Meta executive has denied a rumor that the company trained its new AI models to present well on specific benchmarks while concealing the models’ weaknesses. Ahmad Al-Dahle, VP of generative AI at Meta, stated in a post on X that it’s "simply not true" that Meta trained its Llama 4 Maverick and Llama 4 Scout models on "test sets." This practice could misleadingly inflate a model’s benchmark scores, making it appear more capable than it actually is.

Origin of the Rumor

The rumor appears to have originated from a post on a Chinese social media site from a user claiming to have resigned from Meta in protest over the company’s benchmarking practices. This unsubstantiated claim was further fueled by reports that Maverick and Scout perform poorly on certain tasks. Additionally, Meta’s decision to use an experimental, unreleased version of Maverick to achieve better scores on the benchmark LM Arena has raised concerns among researchers.

Differences in Model Performance

Researchers on X have observed stark differences in the behavior of the publicly downloadable Maverick compared with the model hosted on LM Arena. This inconsistency has led to questions about the accuracy of the models’ benchmark scores.

Acknowledging Quality Issues

Al-Dahle acknowledged that some users are seeing "mixed quality" from Maverick and Scout across the different cloud providers hosting the models. He attributed this to the models being released as soon as they were ready, stating that it will take several days for all public implementations to get "dialled in." Meta will continue to work on bug fixes and onboarding partners to improve the models’ performance.

Conclusion

In conclusion, Meta has denied the rumor that it trained its AI models to present well on specific benchmarks while concealing their weaknesses. The company is working to address the quality issues and inconsistencies in the models’ performance.

FAQs

Q: What is the rumor about Meta’s AI models?

A: The rumor claims that Meta trained its AI models, Llama 4 Maverick and Llama 4 Scout, on "test sets" to artificially inflate their benchmark scores.

Q: Is this rumor true?

A: No, according to Meta executive Ahmad Al-Dahle, the company did not train its AI models on test sets.

Q: Why are some users seeing mixed quality from Maverick and Scout?

A: The inconsistent performance is due to the models being released as soon as they were ready, and it will take several days for all public implementations to get "dialled in."

Q: What is Meta doing to address the quality issues?

A: Meta is working on bug fixes and onboarding partners to improve the models’ performance.

Google Adds Image Search To AI Mode

Jordan Waverly
-
0
Google Adds Image Search To AI Mode

Google AI Mode Now Understands Images

Google AI mode has added a new feature that allows users to upload photos and ask questions about them. This new capability is being rolled out to more users in the US, making it easier to find information and answers using images.

New Image Search Capabilities

Google has added image search capabilities to AI Mode in Search, enabling users to upload pictures and ask questions about them. This feature uses machine learning algorithms to analyze images and provide relevant information. Users can ask questions like “Who is this person?” or “What is this object?” and AI Mode will provide answers based on the image.

How to Use AI Mode

To use AI Mode’s new image search capabilities, follow these steps:

  • Go to Google Search and click on the “AI Mode” button at the top right corner of the page.
  • Click on the “Upload image” button and select a photo from your computer or mobile device.
  • Ask a question about the image, such as “Who is this person?” or “What is this object?”
  • AI Mode will analyze the image and provide relevant information in the search results.
Benefits of AI Mode

AI Mode’s new image search capabilities offer several benefits, including:

  • Improved search results: AI Mode’s machine learning algorithms can analyze images and provide more accurate search results.
  • Increased accessibility: Users with visual impairments or language barriers can use AI Mode to find information and answer questions.
  • Enhanced user experience: AI Mode’s interactive features make it easier for users to find information and engage with the search results.
Conclusion

Google AI Mode’s new image search capabilities are a significant improvement in the search engine’s capabilities. With this feature, users can now upload photos and ask questions about them, making it easier to find information and answer questions. This feature is being rolled out to more users in the US, and we can expect to see even more improvements in the future.

Frequently Asked Questions

Q: Is AI Mode available to everyone?

A: AI Mode is being rolled out to more users in the US, but availability may vary depending on your location.

Q: Can I use AI Mode on my mobile device?

A: Yes, AI Mode is available on both desktop and mobile devices.

Q: Can I ask follow-up questions about an image?

A: Yes, you can ask multiple questions about an image, and AI Mode will provide answers based on the image.

Q: Is AI Mode available in other languages?

A: AI Mode is currently available in English, but Google plans to add support for other languages in the future.

1...899091...741Page 90 of 741

Welcome to MachineDaily.AI, your go-to destination for breaking news, in-depth analysis, and the latest developments in the world of artificial intelligence (AI). Our mission is to provide you with timely and insightful content that explores AI’s vast potential across industries and its growing impact on society. From cutting-edge innovations to ethical debates, MachineDaily.AI is committed to covering AI in all its forms, making complex topics accessible to experts, enthusiasts, and curious minds alike.

Latest

Categories

Useful Links

Our Newsletter

Subscribe Us To Receive Our Latest News Directly In Your Inbox!

© Copyright 2024. All Right Reserved By Machine Daily.