Google has canceled half of the projects in its Area 120 in-house incubator and will shift the division's focus to AI-related projects, TechCrunch reported. The experimental R&D unit will now support AI-first projects that back Google's "deep investment in AI and have the potential to solve important user problems," a spokesperson said. More: - On Tuesday, Google announced that it would eliminate an undisclosed number of positions in Area 120.
- The 14 projects housed in the division have been cut down to seven.
- The canceled projects include a carbon accounting tool, an AR/VR analytics tool, and a map of electric-vehicle charging stations.
- One project that reportedly remains in development is the Aloud app, which transcribes, translates, and narrates videos in multiple languages.
- Employees whose projects were canceled have until the end of January to find new roles within the company or lose their jobs.
Zoom-out: - Area 120 was formed in 2016 to support employees' passion projects within Google.
- Features that have come out of the division include Android smart replies and GameSnacks, an HTML5 gaming platform that's now a part of the Google Ads business unit.
- It also launched AdLingo, a platform that uses AI to create conversational ads.
TECHCRUNCH 
| |
 Dries Depoorter, a Belgian artist, created AI software that identified social media influencers in publicly available camera feeds. Depoorter's project, called "The Follower," highlights how urban surveillance is becoming more possible through facial recognition and other technology. More: - Depoorter initially recorded footage from cameras that capture public spaces like Times Square and Dublin's Temple Bar.
- The open-access cameras frequently livestream on EarthCam and other sites.
- The artist then trained his AI software to scan the footage and identify Instagram influencers with more than 100,000 followers.
- He matched that footage with the images that the influencers took themselves at those landmarks.
- The side-by-side comparisons are available here and show how influencers may spend long time periods trying to find the right pose for a photo.
- Depoorter told Bloomberg, "I show the dangers of technology with my work."
BLOOMBERG
| |
 A message from WIND RIVER Are you making use of all your RTOS's features? Modern real-time operating systems come with updated features that can make development faster and easier.
Join us as we explore the features available in the newest version of VxWorks®, as well as how to take advantage of them to streamline the software development cycle. You will learn: - What features are available in the newest edition of VxWorks
- How to set up and use these new features
- How they can improve the development process for embedded engineers
Register Now!
| |
Chinese search engine gain Baidu released a demo of ERNIE-ViLG, China's most advanced AI image generator that censors politically sensitive words and phrases. The system is being compared to DALL-E 2, Stable Diffusion, and other Western AI image generators. However, ERNIE-ViLG differs in that it understands Chinese prompts, such as traditional Chinese poems, and excels in anime artwork. Background: - Researchers proposed ERNIE-ViLG late last year as part of Baidu's natural-language processing project known as Wenxin.
- ERNIE-ViLG was trained on 145 million Chinese image-text pairs, less than DALL-E 2's 650 million pairs and Stable Diffusion's 2.3 billion pairs.
- Baidu's system does contain 10 billion parameters, more than DALL-E 2 and Stable Diffusion (3.5 billion parameters and 890 million, respectively).
- According to its developers, it achieved state-of-the-art performance in both text-to-image and image-to-text generation.
- Baidu has since released demos on its own platform and the AI community Hugging Face.
Censorship: - ERNIE-ViLG's artworks capture "the cultural specificity of China," according to MIT Technology Review.
- It does block certain words, like political leaders' names, from generating any image outputs. ARS Technica tried out the system and confirmed that it censored "democracy in China," "Chinese flag," and other phrases.
- When they were typed in, the system delivered a warning in Chinese that said, "the input content does not meet the relevant rules..."
MIT TECHNOLOGY REVIEW
| |
 Cruise Origin General Motors' self-driving unit, Cruise, has developed in-house chips for autonomous vehicles, following in the footsteps of Tesla. Cruise plans to deploy the custom chips in its Origin self-driving shuttle by 2025. More: - Switching from third-party Nvidia chips to its own chips is expected to save Cruise money in the long run.
- The company believes it will recoup its investment once it scales up production of its self-driving vehicles.
- According to Reuters, Cruise has developed four chips so far, including a "main brain" chip called Horta and another called Dune for processing sensor data.
- Cruise Origin, announced in 2020, lacks pedals and a steering wheel and has four facing seats that will shuttle multiple passengers.
REUTERS
| |
 A message from SQUARE Take payments anywhere. Sign up for Square and start taking payments right away with a free magstripe reader. No contracts. No monthly fees. Know what you pay. Accept Visa, Mastercard, Discover, and American Express all for one low rate, with no surprise fees. See deposits in your external bank account as soon as the next business day — or instantly in your free Square Checking account. Getting started is simple and free. There are no long-term contracts or surprises—sign up now to start selling in minutes. Sign Up Now 
| |
The U.S. National Institutes of Health (NIH) says it will invest $130M over four years into its new Bridge to Artificial Intelligence program. Called Bridge2AI, the initiative is focused on accelerating AI technology in biomedical and behavioral research. More: - Bridge2AI is aimed at creating best practices and standards for AI-ready datasets.
- The tools could better help researchers prepare data for AI analysis in biomedical research and healthcare.
- A focus is on generating ethically sourced and high-quality datasets, according to NIH acting director Lawrence Tabak.
- As noted by NextGov, data collected in biomedical research and healthcare often lack details or are insufficient.
- AI that is trained on this incomplete data has the potential to generate technologies with bias and other inequities.
- As of today, the NIH has started awarding the funding. It issued four awards for projects that generate AI-usable datasets and three to form a "Bridge Center."
NEXTGOV
| |
🍋 FRESH FUNDING 🍋 - Alcatraz AI (California), provides facial recognition technology for enterprise authentication purposes: $25M A led by Almaz Capital with participation from included EBRD, Endeavor Catalyst, Silverline Capital, Golden Seeds, JCI Ventures, Ray Stata, and LDV Partners.
- Diveplane (North Carolina), helps companies become more transparent by organizing their data with AI: $25M A led by Sigma Defense LLC, Shield Capital, Calibrate Ventures, and L3Harris Technologies.
- Tarci (Israel), a data analytics company that helps small and medium-sized businesses become more efficient: $17M A led by Sound Ventures with participation from Liberty Mutual Strategic Ventures and Global Founders Capital.
- InsideTracker (Massachusets), offers health recommendations based on machine learning, demographic analysis, DNA tracking, and key other health factors: $15M growth led by PeakBridge with participation from Cornucopian Capital, OurCrowd, BASF, and Venture Capital GmbH.
- Awiros (India), a video processing platform that serves as a marketplace for apps that use AI: $7M A led by Inflexor Ventures and Exfinity Venture Partners with participation from VedaVC.
For the latest on AI startup funding and more in the VC community, check out our Inside Venture Capital newsletter.
| |
QUICK HITS - Join ClickUp For Startups to scale success with hands-on support and free resources. Qualifying startups get $3,000 in credit.*
- AI chip startup SambaNova Systems is now shipping DataScale SN30, its second-generation DataScale hardware for AI workloads.
- Intel, Arm, and Nvidia proposed an 8-bit floating point (FP8) standard as a common interchange format for AI, saying it will speed up AI development.
- IT service management company Converge Technology Solutions acquired analytics and AI solutions firm Newcomp Analytics for an undisclosed amount.
*This is a sponsored post.
| |
Upcoming Events: - September 14 - What Do Developers Think of Threat Modeling? (feat. security experts from Security Compass, Dell, and Tricentis) (Register Here) *
- September 14 - AMA with Brian Dean (Founder of Backlinko) (Register Here)
- September 21 - Hired's Navigating Market Uncertainty: The State of Tech Hiring Webinar (Register Here) *
- September 21 - AMA with Kristen Ruby (Founder of Ruby Media Group) (Register Here)
- September 22 - "Top 10 Overlooked VxWorks Features" with Brian Kuhl (Register Here) *
- September 28 - AMA with Leigh-Ann Buchanan (Founder of aīre ventures) (Register Here)
- October 05 - AMA with Ram Bartov (Chief Accounting Officer at TripActions)* (Register Here)
- October 11 - AMA with Mike Malone (Smallstep) (Register Here)
- October 12 - AMA with Bill Glenn (Executive Chairman at Crenshaw Associates) (Register Here)
- October 19 - AMA with Zecca Lehn (Responsibly VC) (Register Here)
- October 25 - Meet Our Fund 4, an Inside.com Summit (Register Here)
*This is a sponsored event.
| |
 |  | Beth is a writer and editor at Inside.com covering artificial intelligence and daily technology news. You can reach her at Beth.Duckett@inside.com. |   | | Editor | Aaron Crutchfield is based in the high desert of California. Over the last two decades, he has spent time writing and editing at various local newspapers and defense contractors in California. When he's not working, he can often be found looking at the latest memes with his kids or working on his 1962 and 1972 Fords. | |
| |
| |
Comments
Post a Comment