EN

KR

UPP News: What's new in the AI field? (Part 14)

UPP News: What's new in the AI field? (Part 14)

Date

October 4th, 2024

Reading Time

5 mins

Subscribe to our newsletter and stay up-to-date with the latest AI news updates.

1. OpenAI’s DevDay brings Realtime API and other treats for AI app developers. 

OpenAI’s DevDay brings Realtime API and other treats for AI app developers
OpenAI’s DevDay brings Realtime API and other treats for AI app developers

October 1st, 2024

Source: OpenAI

OpenAI is making headlines at its 2024 DevDay, announcing new tools like the public beta of the “Realtime API,” which enables low-latency, AI-generated voice responses. Despite recent executive departures, chief product officer Kevin Weil reassured that progress will continue.

The company has reduced API access costs by 99% in response to competitive pressures. The Realtime API allows developers to create near real-time speech interactions using six distinct voices provided by OpenAI, while also integrating with services like Twilio for phone calls, without automatic AI identity disclosure.OpenAI introduced vision fine-tuning, enabling the use of images with text for improved visual tasks, though developers are restricted from uploading copyrighted content. Additionally, a model distillation feature helps optimize smaller AI models for cost efficiency.

Notably, there were no updates on the GPT Store or new AI models during this event, leaving developers awaiting further announcements.

2. Meta confirms it may train its AI on any image you ask Ray-Ban Meta AI to analyze. 

Meta confirms it may train its AI on any image you ask Ray-Ban Meta AI to analyze
Meta confirms it may train its AI on any image you ask Ray-Ban Meta AI to analyze

October 2nd, 2024

Source: TechCrunch

Meta has clarified that images and videos shared with its AI can be used for training purposes, particularly in the U.S. and Canada. While photos and videos captured on Ray-Ban Meta glasses are not automatically used for training unless submitted to AI, once users engage the AI, those images fall under different policies. This means users may unintentionally provide Meta with personal data, raising privacy concerns.

Meta recently introduced new AI features for the Ray-Ban Meta, including live video analysis, which could increase the likelihood of users sharing images for AI training. Their privacy policy states that interactions with AI features can be used to train models, including analysis of shared images.

Meta has faced scrutiny over its use of facial recognition technology and has settled lawsuits regarding data privacy. Users can opt out of having their voice recordings used for AI training when logging into the Ray-Ban Meta app, but many of the privacy issues echo past concerns from the Google Glass era, as smart glasses gain popularity in the tech landscape.

3. Pinterest rolls out genAI tools for product imagery to advertisers. 

Pinterest rolls out genAI tools for product imagery to advertisers
Pinterest rolls out genAI tools for product imagery to advertisers

October 1st, 2024

Source: Pinterest

Pinterest has launched new generative AI features for its advertisers at the Pinterest Presents event, enhancing its Performance+ suite. These tools allow advertisers to transform white or flat backgrounds into lifestyle imagery, a capability now common among major tech companies like Google and Amazon.

One early adopter, Walgreens, tested the technology on 50 products, resulting in a 55% higher clickthrough rate and 13% lower cost-per-click with AI-generated backgrounds compared to standard ones. Pinterest claims that its AI and automation features enable faster campaign creation with 50% less input, leading to a 64% decrease in cost per action and a 30% increase in conversion rates.

Additionally, Pinterest is introducing Promotions tools for advertisers to offer discounts, available in several countries. The Performance+ bidding system will now allow optimization for the highest value rather than just clicks or conversion volume, benefiting global advertisers across various objectives.

4. Runware uses custom hardware and advanced orchestration for fast AI inference. 

Runware uses custom hardware and advanced orchestration for fast AI inference
Runware uses custom hardware and advanced orchestration for fast AI inference

October 1st, 2024

Source: TechCrunch

Runware is an emerging player in the generative AI landscape, quickly impressing users with its fast image generation capabilities—producing images in under a second. The startup has secured $3 million in funding from investors like Andreessen Horowitz and LakeStar, focusing on optimizing its own servers and software to enhance inference speeds.

Instead of renting GPU time like many competitors, Runware offers an image generation API with a cost-per-call fee structure, positioning itself as both cheaper and faster. CEO Flaviu Radulescu emphasizes that their performance advantages stem from their unique orchestration layer and hardware optimizations.

Currently utilizing Nvidia GPUs, Runware aims to develop compatibility with multiple GPU vendors to maintain cost efficiency. By optimizing the entire inference pipeline, the company seeks to build a hybrid cloud infrastructure that could further enhance its competitive edge in AI inference.

5. Microsoft Copilot can now read your screen, think deeply, and speak aloud to you. 

Microsoft Copilot can now read your screen, think deeply, and speak aloud to you
Microsoft Copilot can now read your screen, think deeply, and speak aloud to you

October 1st, 2024

Source: Microsoft

6. Microsoft brings AI-powered overviews to Bing. 

Microsoft brings AI-powered overviews to Bing
Microsoft brings AI-powered overviews to Bing

October 1st, 2024

Source: Microsoft

Microsoft has officially launched Bing generative search, its response to Google’s AI-powered search features, rolling it out to all U.S. users after a pilot phase in July.

Users can activate this feature by searching "Bing generative search" or through specific informational queries. The tool uses AI models to aggregate information from the web, providing summaries and links in response to user queries.Bing generative search aims to improve user intent fulfillment by dynamically matching content and presenting results in an AI-generated format. However, concerns persist about the reliability of AI-generated information, given past incidents where such systems provided misleading or harmful suggestions.

While Microsoft claims that Bing generative search does not diminish traffic to publisher sites, previous studies indicate that Google’s similar AI Overviews may have negatively impacted publisher traffic by de-emphasizing article links. With Google holding a dominant 81.95% of the global search market compared to Bing's 10.51%, any changes to Bing's search experience may have limited overall impact.

Newsletter

DISCOVER MORE

LET’S TALK...

Content delivered to your inbox

ENTER YOUR EMAIL

YOU WANT TO...

Subscribe
KSA Cloud
ISO 9001:2015
ISO 27001:2022

Hanoi, Vietnam

Web3 Tower, No. 15, Alley 4, Duy Tan, Cau Giay, Hanoi, Vietnam

© 2025 UPP Global Technology JSC

Look up for solutions? Look for UPP!

PRIVACY POLICY