OpenAI Unveils Voice Engine

PLUS: NYC's AI Chatbot Under Fire for Illegal Advice, OpenAI Enhances ChatGPT with Prominent Source Links and more.

Sponsored by

Artificial Intelligence online short course from MIT

Study artificial intelligence and gain the knowledge to support its integration into your organization. If you're looking to gain a competitive edge in today's business world, then this artificial intelligence online course may be the perfect option for you.

  • Key AI management and leadership insights to support informed, strategic decision making.

  • A practical grounding in AI and its business applications, helping you to transform your organization into a future-forward business.

  • A road map for the strategic implementation of AI technologies in a business context.

Today:

  • OpenAI Unveils Voice Engine

  • Elon Musk Unveils Grok-1.5 Chatbot

  • Scale AI Valuation Soars in Accel-Led Round

  • NYC's AI Chatbot Under Fire for Illegal Advice

  • OpenAI Enhances ChatGPT with Prominent Source Links

OpenAI reveals artificial intelligence tool to re-create human voices

OpenAI introduces Voice Engine, a tool that replicates human voices from brief recordings, currently being tested by select businesses. With caution due to potential misuse, the program synthesizes natural-sounding voices resembling the original speaker, capable of reading texts in various languages. 

Demonstrated examples include translating an English speaker's voice into multiple languages while maintaining their accent. Applications range from aiding nonreaders to providing translation services. OpenAI emphasizes responsible deployment and societal adaptation to this technology, considering risks such as fraud and misinformation. While not widely released yet, similar options exist, prompting discussions on safeguards against abuse.

Musk's Grok-1.5 AI chatbot to be available next week

Elon Musk's AI startup xAI will roll out Grok-1.5 chatbot on his social media platform X next week, aiming to surpass current AI standards. This upgraded version of Grok will be accessible to early testers and existing users. 

Musk's move comes after announcing plans to open-source Grok, following a legal dispute with Microsoft-backed OpenAI over its mission shift towards profit. With xAI competing against industry giants like Google, Musk's strategic push signifies the evolving landscape of AI technology. 

Scale AI Nears $13 Billion Valuation in Accel-Led Round

Venture capital firm Accel is poised to lead a funding round for Scale AI, propelling its valuation to around $13 billion, an 80% increase. Scale AI, known for data labeling services catering to AI startups like OpenAI, aims to secure hundreds of millions in this round. 

With its expertise in labeling images and text for AI models, Scale AI continues to attract significant investment, highlighting its pivotal role in the AI ecosystem's growth. 

NYC’s AI Chatbot Tells Businesses to Break the Law

New York City’s AI chatbot, part of Mayor Eric Adams' initiative to modernize city services, is advising businesses to break the law. Launched in October 2023 to offer guidance on starting and running businesses, the bot often provides incomplete or inaccurate information. For instance, it falsely states that landlords can discriminate based on income sources, encourages businesses to disregard consumer and worker protections, and even suggests illegal practices like withholding funeral service prices. 

Despite its pedigree using Microsoft's Azure AI services, the bot's flaws highlight broader issues with relying on AI for critical information dissemination, raising concerns about its impact on small businesses and individuals.

OpenAI makes source links more prominent to people who are searching with ChatGPT.

OpenAI spruces up ChatGPT by highlighting source links, a change that jazzes up responses. Previously buried in quotes, links now pop as hyperlinked words in the bot's answers, mimicking rivals like Perplexity. 

This tweak boosts clarity, aiding users in pinpointing references effortlessly. Exclusive to subscribers of ChatGPT Plus, Team, and Enterprise, this upgrade aims to enhance the browsing experience for paying customers.

🧠RESEARCH

Garment3DGen is a new tool for creating 3D clothing from a single image, usable with real or computer-generated images. It employs advanced techniques to deform a base mesh into the desired shape and texture, allowing for realistic garment simulation without manual intervention. The method offers flexibility and accuracy for diverse garment designs.

Gamba combines Gaussian splatting and Mamba to efficiently reconstruct 3D assets from single images, addressing limitations of previous methods like lengthy optimization and high memory usage. It leverages a large number of 3D Gaussians and a sequential network for context-dependent reasoning, achieving competitive results in speed and quality on real-world datasets.

EgoLifter is introduced as a system for segmenting scenes captured from egocentric sensors into individual 3D objects. Utilizing 3D Gaussians as the representation and Segment Anything Model (SAM) masks for weak supervision, it offers flexible object definitions. A transient prediction module filters dynamic objects, leading to automatic, high-performance 3D segmentation in egocentric scenarios, demonstrated on the Aria Digital Twin dataset.

sDPO, an extension of Direct Preference Optimization (DPO), is introduced to align large language models (LLMs) with human preferences more effectively. By utilizing preference datasets incrementally instead of all at once, it enables better alignment and improves model performance, surpassing even larger LLMs in certain cases.

LITA (Language Instructed Temporal-Localization Assistant) addresses the temporal localization gap in video processing by introducing time tokens for better representation, SlowFast tokens for fine temporal resolution, and emphasizing temporal localization data. It introduces the Reasoning Temporal Localization (RTL) task and dataset, showcasing significant performance improvements and enhancing video-based text generation.

🛠️TOP TOOLS

Sonauto - Turn prompts, lyrics, and melodies into full songs.

Amplemarket – AI platform for modern sales teams

Glimmer - AI search for large PDFs

Swsh - Easiest way to share photo albums

📲SOCIAL MEDIA

🗞️MORE NEWS

SambaNova announces new AI Samba-CoE v0.2 that already beats Databricks DBRX

SambaNova Systems introduces Samba-CoE v0.2, a high-speed AI model surpassing rivals like Databricks' DBRX. It operates at 330 tokens per second, maintaining precision with just 8 sockets. Leveraging open-source models, it promises efficiency and hints at future advancements with v0.3. VENTUREBEAT

Google DeepMind CEO Demis Hassabis gets UK knighthood for ‘services to artificial intelligence’

Demis Hassabis, CEO of DeepMind, receives a knighthood in the U.K. for his contributions to artificial intelligence. Recognized alongside other AI figures, his achievements include mastering chess as a teenager and developing AlphaGO. This honor highlights the U.K.'s push to lead in AI innovation, with DeepMind as a prominent export. TECHCRUNCH

Congress bans staff use of Microsoft's AI Copilot

The U.S. House imposes a strict ban on congressional staffers' use of Microsoft Copilot due to cybersecurity concerns over data leakage. This move reflects the government's efforts to regulate AI internally while navigating its potential risks. Microsoft aims to address these concerns with government-specific tools. AXIOS

AI tool predicts lethal heart rhythm with 80% accuracy

A Leicester study reveals an AI tool accurately predicts lethal heart rhythm 80% of the time. Led by Dr. Joseph Barker and Professor Andre Ng, the tool, VA-ResNet-50, analyzed 270 adults' Holter ECGs, aiding in early detection of ventricular arrhythmia, potentially saving lives. The findings challenge current clinical guidelines, emphasizing AI's role in improving patient risk assessment. MEDICAL LIFE SCIENCES

Chaotic departure of StabilityAI chief raises doubts over start-up’s future

The future of StabilityAI, an AI start-up valued at $1bn, is uncertain following the departure of its founder Emad Mostaque. Mostaque resigned as CEO and handed over voting control to other board members, citing a desire to prioritize open source and the benefit of humanity. However, former executives and investors accuse Mostaque of being an unreliable leader who failed to monetize the company's products and face mounting pressure from financial backers. StabilityAI is seeking a new CEO to achieve its first profit and has faced legal battles and lawsuits from Getty Images and artists over the use of copyrighted material. FINANCIAL TIMES

AI chatbots are improving at an even faster rate than computer chips

AI chatbots are advancing faster than computer chips, with large language models (LLMs) improving rapidly. After eight months, LLMs require only half the computing power to achieve the same benchmark score, outpacing the rate of hardware improvement. This suggests AI developers are becoming more efficient in utilizing resources. NEWSCIENTIST

The Gaming Industry’s Edge in the Artificial Intelligence Revolution

The gaming industry pioneers AI integration, enhancing player experiences with immersive environments and lifelike NPCs. AI not only refines dialogue but also manages data for expansive digital realms. Gaming AI, akin to Amazon and Netflix, utilizes player behavior for development and business insights, heralding an AI-driven future across industries. FORBES

Surprise! Google will let the Pixel 8 run on-device AI after all

Google relents under public pressure, allowing Pixel 8 to utilize on-device AI, matching Pixel 8 Pro's capabilities. Previously cited "hardware limitations" debunked, as Samsung implements similar features in older models. Allison Johnson celebrates community triumph, affirming peer pressure's efficacy in tech advancements. Victory for Pixel enthusiasts. THE VERGE

What'd you think of today's edition?

Login or Subscribe to participate in polls.

What are MOST interested in learning about AI?

What stories or resources will be most interesting for you to hear about?

Login or Subscribe to participate in polls.

Reply

or to participate.