Welcome to Daily Zaps — we’re like the Dunkin Donuts of AI newsletters, the world runs on Daily Zaps. 🏃‍♂️

Google Upgrades Search Generative Experience

Google has unveiled significant enhancements to its AI-powered tool, the Google Search Generative Experience (SGE), including the introduction of the "SGE while browsing" Search Lab feature that extends SGE's functionality beyond traditional search pages. The upgrades encompass the incorporation of definitions within AI-generated search responses, offering comprehensive contextual information, and a boost for programmers through Bard-inspired code debugging features. 

While these updates are being gradually rolled out, users can sign up for Google Search Labs to access these AI features. Notably, SGE's most groundbreaking feature expansion enables it to accompany users during internet browsing, generating article summaries and facilitating deeper exploration of web pages, accessible on iOS, Android, and desktop. Additionally, an upcoming tool will allow users to preview related definitions and contextual elements by hovering over terms, benefiting researchers, although it's advised to verify AI-generated data through conventional means due to potential inaccuracies.

GPT-4 Can Now Replace Human Moderators

OpenAI is confidently addressing a formidable challenge in the tech industry: scalable content moderation, asserting that its GPT-4 technology could potentially replace tens of thousands of human moderators with nearly equivalent accuracy and greater consistency. OpenAI has already employed GPT-4 to enhance its content policies, content labeling, and decision-making processes.

The technology offers three key advantages over traditional methods. Firstly, it ensures consistent judgment by avoiding varied human interpretations of policies, which can be complex and constantly evolving. Secondly, GPT-4 accelerates policy development, shortening the typically time-intensive drafting, feedback, and refinement stages from weeks to hours.

Thirdly, the well-being of human moderators exposed to distressing content is improved, contrasting with the often minimal mental health support available to these workers. While major platforms like Meta, Google, and TikTok rely on machine moderation, OpenAI acknowledges its dependence on clickworkers for annotation and labeling, highlighting potential contradictions in its approach. Despite these advancements, the challenge of accurately moderating ambiguous, misleading, or controversial content remains, potentially exacerbated by generative AI technologies' potential to produce misinformation.

Nevertheless, OpenAI's pursuit of AI-driven content moderation seeks to alleviate a longstanding issue in the online realm while recognizing the need for continued refinement and collaborative efforts to address its limitations.

Together With Our Partners

Score Those Straight A's with Dropchat: Your AI Study Buddy

Dropchat Flashcard Demo Video

Ever wished you had an AI sidekick to help you secure those perfect grades?

Meet Dropchat — your go-to app for turbocharging your academic game and nailing those straight A’s.

Here’s how it works:

• 📖 Toss in your class notes, PDFs, or even a YouTube video, and watch as Dropchat works its magic and turns your content into a study sidekick.

• 💬 Didn’t read the material? No problem, chat it up and Dropchat will give you all the answers.

• 📝 Want to challenge yourself? Create interactive flashcards for a self-quiz session.

• 🔗 Share the chatbot via code snippet or URL with your fellow students

Wave goodbye to study stress, and say hello to more free time and top-notch grades — all yours with Dropchat. Ready to conquer your courses? Dive into Dropchat for free today!", Get started with Dropchat for free.

AI Camera Catches 300 UK Law Breakers

In the UK, an artificial intelligence camera system has successfully identified nearly 300 instances of law-breaking by drivers within its initial three days of operation. These motorists were observed either using mobile phones while driving or not wearing seatbelts. Positioned along the A30 near Launceston, Cornwall, this stand-alone AI road safety camera employs multiple high-speed cameras, infra-red flash, and advanced lensing and filtering technology to capture clear images of passing vehicles. Despite the AI's ability to detect potential violations, all images are subsequently reviewed by human personnel, as clarified by Devon and Cornwall Police.

Any images flagged for potential offenses are anonymized and then sent for human confirmation of the violation. If a violation is confirmed, the driver may receive either a warning letter or a notice of intended prosecution, depending on the severity of the infraction. The system detected a total of 590 seatbelt-related and 45 mobile phone-related offenses across different roads in Devon and Cornwall. This initiative is part of ongoing efforts to enhance road safety and reduce fatalities in the region, with the system aimed at curbing risky behaviors like mobile phone use and seatbelt non-compliance.

Saudia Arabia Desperate For NVIDIA H100 Chips

Nvidia's graphics processing units (GPUs) are experiencing a surge in demand, with Saudi Arabia and the United Arab Emirates (UAE) acquiring a substantial number of chips for the development of artificial intelligence software, as reported by the Financial Times, citing anonymous sources. Saudi Arabia has acquired a minimum of 3,000 Nvidia H100 chips, while the UAE, in its pursuit of building its own significant language model named Falcon, has also secured access to a significant quantity of Nvidia chips.

Although Nvidia has not yet responded to queries for comment, the company's GPUs are evidently becoming instrumental in supporting AI advancements in the region. This development coincides with Nvidia Corporation witnessing a notable increase of over 7% in its stock value, attributed in part to Morgan Stanley's endorsement of the recent dip in the company's stock as a favorable buying opportunity.

Midjourney Releases Cluster Upgrade

Midjourney is introducing a significant enhancement to its GPU cluster, poised to benefit numerous users, particularly those subscribed to the Pro and Mega tiers. This upgrade promises substantial acceleration in image generation times, with an /imagine prompt that previously took around 50 seconds now expected to complete in just 30 seconds, marking a 1.5x speed increase.

The augmentation is made possible by the expansion of Midjourney's backend GPU capacity, which enables AI models to process prompts more rapidly through parallel processing. The enhanced efficiency not only translates to quicker responses for Pro and Mega users but also results in cost savings, estimating a 1.5x reduction in image generation expenses for these tiers.

Additionally, any surplus GPU capacity stemming from the upgrade will be utilized to intermittently hasten image generation for other tiers, including Free and Basic subscribers, extending the potential benefits of speed improvement across various subscription levels, including Fast, Relax, and Turbo Modes. This cluster upgrade marks a significant stride forward for Midjourney, enhancing both the speed and value of its services for a broad range of users.