A naukri.com initiative
Google News
TechCrunch
3w
83
Image Credit: TechCrunch
Google says its new AI models can identify emotions — and that has experts worried
- Google has unveiled the PaliGemma 2 model family, a system that includes the ability to “identify” emotions in photos using closed-caption features.
- The PaliGemma system has to be fine-tuned for emotion detection, and industry experts including a professor in data ethics at the Oxford Internet Institute raised concerns about an openly available emotion detector.
- Most emotion detectors utilise the work carried out by the psychologist Paul Ekman, however, his theory that humans share six emotions has been called into question, as it is now recognised that different cultures and backgrounds have different ways of expressing themselves.
- Experts believe that while detecting generic signifiers in emotion is possible, it is not something that can be fully “solved” due to the inherent complexity of emotions.
- Emotion detection tends to be unreliable and biased by the assumptions of the designers and AI systems. For example, a study showed models could develop unintended preferences for certain expressions, and more recent work suggests that emotional analysis models assign more negative emotions to black people’s faces than white people’s faces.
- Emotion detection systems have been subject to oversight. The EU’s AI Act prohibits schools and employers from deploying the technology in high-risk contexts.
- Researchers have criticised Google's benchmark, 'FairFace', for being a biased metric that only represents a handful of race groups.
- Chief AI scientist at AI Now Institute Heidy Khlaaf said, “We cannot infer emotions from facial features alone”.
- The release of open models like PaliGemma could lead to misuse or abuse that could discriminate against marginalised groups in law enforcement and human resources.
- Responsible innovation needs to factor in the potential consequences from product development and throughout the lifecycle thereof, says Sandra Wachter, a professor in data ethics and AI at the Oxford Internet Institute.
Read Full Article
5 Likes
Androidauthority
3w
227
Image Credit: Androidauthority
Live Captions on Android can now help you read between the lines
- Google is enhancing Live Caption with Expressive Captions on Android.
- Expressive Captions use AI to recognize tone, volume, and ambient sounds and convey them with stylized captions and labels.
- The feature is available in the US for English on devices running Android 14 and above.
- Google introduced Live Caption as a feature back in Google I/O 2019 and is now upgrading it with Expressive Captions.
Read Full Article
13 Likes
Droid-Life
3w
148
Image Credit: Droid-Life
December Android Feature Drop Brings Expressive Captions, Quick Share QR Codes, More
- Google's Android-wide Feature Drop for December brings new features including Gemini extensions, expressive captions, Gemini in Lookout, Quick Share QR Codes, Google Drive auto enhancements, and Emoji Kitchen update.
- Gemini extensions allow users to access services like Utilities, Spotify, Messaging, Calling, and Maps. Expressive captions enhance the intensity of emotion in captions and are applicable to live and social content.
- Gemini in Lookout utilizes Gemini 1.5 Pro to provide accurate and detailed image descriptions in Google's Lookout app.
- Quick Share QR Codes enable fast and easy media transfer between devices without the need for device verification or changing sharing settings.
Read Full Article
8 Likes
Androidauthority
3w
52
Image Credit: Androidauthority
Android’s Quick Share will now let you transfer photos and videos using QR codes
- Quick Share on Android now allows file transfer using QR codes.
- The feature eliminates the need to add contacts or change sharing settings.
- Quick Share is a feature borrowed from Samsung and now introduced by Google.
- Users can select a media file, generate a QR code, and have others scan it for secure file transfer.
Read Full Article
3 Likes
Discover more
Droid-Life
3w
271
Image Credit: Droid-Life
December 2024 Android Security Update Available for Google Pixel Devices
- Google released the December Pixel update, the stable version of Android 15 QPR1.
- The update includes the December Pixel Feature Drop with 20+ changes.
- Rollout for the update has begun for various Pixel devices.
- Users can check for the update or manually update using factory images or OTA files.
Read Full Article
16 Likes
Androidauthority
3w
288
Image Credit: Androidauthority
The latest Gemini Extensions are ready to search Spotify lyrics, make calls for you, and much more
- Google formally welcomes Gemini Extensions for Spotify, Utilities, Messages, and Phone.
- The big news is Spotify getting its own Gemini Extension, allowing users to search for lyrics, find playlists, and play specific songs or artists.
- Google also introduces Extensions for Utilities, Messages, and Phone, offering a range of features including controlling timers and alarms, sending texts, and making conversational calls.
- Google hints that the Home Extension for Gemini could soon leave preview and provides more information about places when using Gemini with Maps.
Read Full Article
17 Likes
Droid-Life
3w
433
Image Credit: Droid-Life
December Pixel Feature Drop is Here: 20+ New Features and Changes Included
- Google released December Pixel update as December Pixel Feature Drop, it includes at least 25 changes on Pixel 6 and new Pixel 9 & 9 Pro series.
- Expressive Captions is a Pixel 6+ feature that adds intensity and emotion to captions while watching videos.
- Gemini AI has access to more apps including Spotify and Messages; its saves user preference to provide better and tailored responses.
- Google Recorder introduces Clear Voice Mode for Pixel 9 series to reduce noise and focus on speakers.
- Pixel Screenshots app on Pixel 9 series adopts new features like automatic categories and search filters for easier organization.
- Gboard will feature Pixel Studio stickers to share in Pixel 9 conversations on social media.
- Quick Share feature adds an easier process to share via QR codes, and Now Playing feature shows album art for each song in the Pixel 6+.
- Pixel Tablet gets a built-in Google VPN feature and Lockscreen widgets, while Pixel Watch gets expansion and few new features.
- Identity Check requires face or fingerprint authentication before changing sensitive settings in case the device is stolen or forcefully moved.
- Simple View is a new feature that aims to navigate the device in an easier way, increasing font size and touch sensitivity.
Read Full Article
26 Likes
Androidauthority
3w
70
Image Credit: Androidauthority
Call Screen on Pixel devices gets Gemini-powered contextual replies
- Call Screen on Pixel devices is getting an upgrade with the December 2024 Pixel Drop.
- The feature will now suggest better contextual replies using Gemini Nano.
- This AI-powered feature allows users to answer or ask follow-up questions without answering the call.
- The update also brings improvements like Expressive Captions and Automatic Call Screening view.
Read Full Article
4 Likes
TechCrunch
3w
26
Image Credit: TechCrunch
Google’s latest feature drop includes new Gemini extensions and accessibility features
- Google has introduced new updates for Pixel and Android devices as part of the December feature drop.
- New Gemini extensions have been added, allowing users to explore songs, send messages, make calls, and control smart home devices through Gemini.
- Accessibility enhancements include image Q&As for visually impaired users, expressive captions to reflect emotions on the screen, and better auto-enhancements in Google Drive.
- Pixel devices receive updates such as an easy-to-operate mode, theft protection-related updates, improved call screening, and enhancements to the Pixel Screenshots app.
Read Full Article
1 Like
Cloudblog
3w
39
Image Credit: Cloudblog
How Current leveraged Spanner to build a resilient platform for banking services
- Current, a fintech start-up, utilized Spanner, Google’s globally distributed database, to create a scalable and robust technological foundation for its financial services.
- Central to the fintech’s success was a cloud-native infrastructure built on Google Cloud although its previous self-hosted graph database lacked cloud-native capabilities and horizontal scalability.
- The cloud-native platform required a database solution with consistent writes, horizontal scalability, low read latency under load, and multi-region failover.
- Spanner emerged as the ideal solution for the fintech start-up, fulfilling all its requirements.
- Current previously faced a challenge of making financial services available to low-income Americans without a bank account, solving which they developed a modern core banking system to power its platform.
- Their Spanner-powered user graph service delivered the consistency, reliability, and scalability essential for their financial platform.
- It dramatically improved resilience, reduced RTO and RPO by more than 10x, and cut times to just one hour implementing streamlined data restoration process.
- The company has reduced its availability-related incidents to zero since migrating to Spanner, eliminated any potential disruption in customer satisfaction and potential churn.
- Current’s Spanner-powered infrastructure empowers it to deliver reliable and efficient financial services, critical for building and maintaining member trust.
- Their strategic migration process to Spanner resulted in a zero-downtime, zero-loss cutover, minimizing risks and maximizing efficiency.
Read Full Article
2 Likes
Cloudblog
3w
122
Image Credit: Cloudblog
Moloco: 10x faster model training times with TPUs on Google Kubernetes Engine
- Moloco offers AI-powered advertising solutions that drive user acquisition, retention, and monetization efforts.
- To address the challenges, companies are seeking data-driven approaches to enhance their advertising effectiveness.
- Moloco leverages predictions from a dozen deep neural networks to deliver higher conversion rates and return on investment.
- Moloco's rapid growth required an infrastructure that could handle massive data processing and real-time ML predictions while remaining cost effective.
- Google Cloud, Moloco's infrastructure provider, aligned with its rapidly growing data and machine learning workloads to optimize customer advertising performance.
- Google Kubernetes Engine (GKE) was a primary reason for Moloco selecting Google Cloud over other cloud providers.
- Moloco leveraged GKE's AI and ML capabilities to optimize the management of its compute resources, minimizing idle time, generating cost savings while improving performance.
- Training complex ML models with GKE and Google's specialized processing units for machine learning workloads, resulted in significant enhancements in training speed and efficiency.
- In addition to training models on TPUs, Moloco also uses GPUs on GKE to deploy ML models into production.
- By harnessing Google Cloud's solutions, Moloco was able to scale its infrastructure, accelerate its ML training, and deliver exceptional ad experiences to its customers.
Read Full Article
7 Likes
Cloudblog
3w
166
Image Credit: Cloudblog
Registration is open for Google Cloud Next 2025
- Registration is now open for Google Cloud Next 2025 in Las Vegas, which will be held from April 9-11, 2025. The conference is expected to be even bigger and better than previous years, with hands-on experiences, inspiring content, and problem-solving with top partners.
- Attendees will have the opportunity to learn from top experts and peers in the field, gaining new ideas, breakthrough skills, and actionable knowledge exclusive to Google Cloud Next 2025.
- The conference will feature AI-focused experiences, allowing participants to explore the latest technology, build agents, and learn how AI is being used to drive business growth.
- Additionally, Next provides a platform for networking with industry experts, peers, and influential minds in technology, fostering collaborations and shaping the future of various industries.
Read Full Article
10 Likes
Cloudblog
3w
0
Image Credit: Cloudblog
How the Air Force Research Laboratory is Advancing Defense Research with AI
- The U.S. Air Force Research Laboratory (AFRL) is collaborating with Google Cloud to leverage artificial intelligence (AI) and machine learning (ML) capabilities to drive innovation and accelerate research in various domains.
- Google Cloud’s technology will propel research and development for the U.S. Air Force and Space Force, aligning with broader Department of Defense (DoD) endeavours to integrate AI into critical operations and bolster national security.
- By harnessing Google Cloud’s scalable infrastructure, the AFRL is driving innovation and ensuring the U.S. Air Force and Space Force remain at the forefront of technological advancement.
- With Google Cloud’s infrastructure, the AFRL's bioinformatics data pipeline automation paved the way for further exploration and expansion on a use case that had never been done before.
- The AFRL built an efficient, custom web application, based on Google App Engine, to ease maintenance for non-programmer researchers and lab staff and capture usage metrics for future hardware investment decisions.
- The FOCUS Mission Readiness App built on Google Cloud utilizes various infrastructure services to collect and analyze real-time data from wearables, providing personalized insights and recommendations for fatigue interventions and predictions.
- The AFRL leveraged Google Cloud’s Vertex AI and Jupyter Notebooks, through Workbench, Compute Engine, Cloud Shell, Cloud Build and more to get a head start in creating a pipeline that could be used for sharing, ingesting, and cleaning their code.
- The AFRL is successfully driving innovation and improving their research capabilities with Google Cloud’s comprehensive generative AI offerings and collaborative environment.
- The AFRL's collaboration with Google Cloud is a great example of how AI and cloud services can be used as a driving force behind innovation, efficiency, and problem-solving across agencies.
- This collaboration will be crucial for unlocking the full potential of AI and cloud computing, ensuring that agencies across the federal landscape can leverage these transformative technologies for improved efficiency, effectiveness, and security.
Read Full Article
Like
Analyticsindiamag
3w
74
Image Credit: Analyticsindiamag
Google Unveils PaliGemma 2 Vision-Language Models for Advanced Task Transfer
- Google has introduced PaliGemma 2, a family of vision-language models for advanced task transfer.
- The models come in three sizes and three resolutions, allowing for transfer learning across diverse domains.
- PaliGemma 2 achieved state-of-the-art results in various fields, including molecular structure recognition and table structure analysis.
- The models are designed for accessibility and can operate on low-precision formats for on-device inference.
Read Full Article
4 Likes
Digitaltrends
3w
231
Image Credit: Digitaltrends
The Google Pixel 9 may soon get a powerful phone call feature
- Google Pixel 9 series offers Call Notes feature in Google Phone app.
- An APK teardown suggests a new feature called 'Call To-do List' may be available soon.
- The new feature will use AI to summarize actionable to-dos from phone calls.
- Google is continuously enhancing phone features through AI integration.
Read Full Article
13 Likes
For uninterrupted reading, download the app