🗞 OpenAI Empowers Businesses with GPT-4 Customization 🤖

🔎 The Latest on the AI Frontier:

• OpenAI Empowers Businesses with GPT-4 Customization 🤖

• Condé Nast and OpenAI Forge Content Partnership 📰

• Microsoft Unveils Powerful Phi-3.5 AI Models 🧠

• Google AI Overviews: Minimal Impact on Publisher Traffic 🔍

• Nvidia Announces New RTX 4070 with GDDR6 Memory 🖥️

• AI Voice Cloning Scams on the Rise: Consumer Alert 🎙️

• Best Buy Launches AI-Powered Delivery Tracking System 📦

🤖 OpenAI allows companies to fine-tune GPT-4 model, shifting towards AI as a service. Link

• OpenAI is offering companies the ability to customize its ChatGPT-4 model, with 1 million free training tokens per day until September 23rd.

• This move enables businesses to optimize AI for specific needs, potentially improving performance in areas like coding, writing, and customer service.

• OpenAI has also signed content agreements with major publishers like Condé Nast and Time to train its models on their content libraries, signaling a shift in how AI platforms monetize media.

📰 OpenAI signs multi-year deal with Condé Nast to train on and surface their content. Link

• OpenAI will have permission to train its AI models on content from Condé Nast publications like The New Yorker, Vogue, and Wired.

• The partnership will allow OpenAI to surface Condé Nast stories in ChatGPT and its search prototype SearchGPT.

• This adds to OpenAI's growing list of publisher partnerships, which includes Associated Press, Axel Springer, The Atlantic, and others, as the company seeks to expand its training data.

🧠 Microsoft releases powerful new Phi-3.5 AI models, outperforming competitors. Link

• Microsoft unveiled 3 new open-source AI models: Phi-3.5-mini-instruct (3.82B parameters), Phi-3.5-MoE-instruct (41.9B parameters), and Phi-3.5-vision-instruct (4.15B parameters).

• The models beat competitors like Google's Gemini 1.5 Flash and OpenAI's GPT-4o on some benchmarks, while being freely available under an MIT license.

• Phi-3.5 models excel at tasks like code generation, math problem-solving, and multimodal reasoning with images/video.

🔍 Google's AI Overviews haven't significantly impacted publisher traffic, reports say. Link

• Major publishers Dotdash Meredith and Ziff Davis report "negligible" impact on their web traffic from Google AI Overviews.

• AI Overviews appear on about 8-15% of relevant searches for these publishers, with minimal effect on click-through rates so far.

• Google is expanding AI Overviews to more countries and adding new features like in-text links to make it easier for users to visit source websites.

🖥️ Nvidia announces new RTX 4070 variant with GDDR6 memory. Link

• New version uses slower GDDR6 memory instead of GDDR6X, reducing bandwidth from 504 GB/s to 480 GB/s.

• Nvidia claims performance will be "similar" to the original GDDR6X model despite the change.

• New cards will be available from partners in September, likely to improve supply and availability.

🎙️ AI-powered voice cloning scams surge, posing new risks for consumers. Link

• Deloitte predicts AI-enabled fraud losses could reach $40 billion in the U.S. by 2027, up from $12.3 billion in 2023.

• Cybersecurity experts warn that AI can clone voices in as little as 30 seconds, potentially tricking voice recognition systems used by banks.

• The BBB recommends verifying callers' identities, using multifactor authentication, and educating family and friends about the risks of AI voice scams.

📦 Best Buy unveils AI-powered delivery tracking, revolutionizing retail logistics. Link

• New system offers minute-by-minute updates, addressing consumer demand for transparency with 30% of global consumers open to AI-driven order updates.

• 80% of consumers prefer at least four delivery updates per order, while two-thirds of retailers fail to provide basic "shipped" notifications.

• AI-powered tracking could influence customer behavior, potentially encouraging larger online purchases and transforming last-mile delivery efficiency.

Have any feedback? Send us an email