AI Spotlight keeps you up-to-speed on the latest cool stuff in AI and tech. This week, we'll show you new tools, how businesses can use AI, and what's hot in the world of AI.

In today’s email:

• Meta Releases AI Model That Can Check Other AI Models' Work

• Midjourney to Introduce AI Image Editing Tool for Web Users

• How AI-generated content is upping the workload for Wikipedia editors

• How to run Meta’s Llama 3.2 AI model locally on your phone

• 📢Top AI Tools of the Week

Meta Releases AI Model That Can Check Other AI Models' Work

Meta, the parent company of Facebook, has announced a significant breakthrough in AI technology: a "Self-Taught Evaluator" that can assess the performance of other AI models without human involvement. This tool, part of a batch of new AI models from Meta's research division, uses a "chain of thought" process similar to OpenAI's latest models to break down complex tasks into manageable steps, boosting accuracy. Meta's development could pave the way for AI systems that learn from their own mistakes, potentially eliminating the need for costly human feedback loops in AI training.

Key Points

1. New AI Model: Meta's "Self-Taught Evaluator" can assess other AI models' outputs without human input, marking a shift toward autonomous AI systems.

2. Chain of Thought Technique: The model uses a structured reasoning process to tackle complex problems in fields like coding, math, and science, improving accuracy.

3. Fully AI-Generated Training Data: Meta's researchers trained the evaluator using data created entirely by AI, eliminating the need for human involvement at this stage.

4. Impact on AI Development: The model could reduce the reliance on Reinforcement Learning from Human Feedback (RLHF), a labor-intensive process currently used to improve AI performance.

5. Future Implications: Self-evaluating AI systems may lead to digital assistants capable of completing tasks independently, opening new possibilities for autonomous AI applications.

6. Competition: While Meta has made its research publicly available, competitors like Google and Anthropic have conducted similar research but are less open about releasing their models.

Source: https://www.gadgets360.com/ai/news/meta-ai-release-check-other-models-work-6823539

Midjourney to Introduce AI Image Editing Tool for Web Users

Midjourney is preparing to launch a new web tool that will allow users to edit any uploaded images using its generative AI technology. This upgraded feature will enable users to retexture objects and alter the colors and details of images through simple text-based commands. While this development promises to make image editing more accessible, it also raises concerns about potential misuse, particularly in relation to deepfakes and copyright infringement. As AI-generated and AI-modified images become increasingly common, questions surrounding their ethical implications continue to grow.

Key Points:

• New Feature: Midjourney is set to release a tool that allows users to edit and modify images using generative AI.

• Launch Timeline: The tool is expected to be available “early next week,” initially limited to a subset of users for testing.

• Functionality: Users can adjust textures, colors, and image details based on text instructions.

• Ethical Considerations: The rise of AI-edited images complicates the ability to distinguish between AI-modified and AI-generated content, sparking debates about transparency and labeling.

• Deepfake Concerns: Deepfakes are rapidly increasing, with a 900% rise in 2023 alone, heightening concerns about disinformation and misuse.

• Regulation: Legal frameworks, such as California’s stalled deepfake law, aim to address the challenges of AI-aided impersonation and content manipulation.

• Safety Measures: Midjourney plans to use both human and AI moderators to monitor the tool's use and prevent abuse, though the company is still refining its approach to user access and safeguards.

Source: https://techcrunch.com/2024/10/19/midjourney-plans-to-let-anyone-on-the-web-edit-images-with-ai/

Former OpenAI technology chief Mira Murati to raise capital for new AI startup, sources say

Mira Murati, the former chief technology officer at OpenAI, is reportedly raising funds for a new AI startup, according to sources familiar with the matter. Known for her pivotal role in developing groundbreaking technologies like ChatGPT and DALL-E, Murati’s new venture is expected to focus on building AI products using proprietary models. With her reputation in the field, the startup could raise over $100 million, although details remain under wraps. Murati is also said to be recruiting former OpenAI employees, and notable researcher Barret Zoph may join her efforts.

Key Points

• New Venture: Mira Murati, former OpenAI CTO, is in early talks with venture capitalists to raise over $100 million for her new AI startup.

• Focus: The startup will likely develop AI products based on proprietary models, though Murati's exact role, such as CEO, remains unconfirmed.

• Recruitment: Murati has reportedly been recruiting ex-OpenAI employees, and Barret Zoph, another prominent researcher, may become involved.

• Previous Achievements: Murati played a key role in leading projects like ChatGPT, DALL-E, and the GPT-4 model during her six years at OpenAI.

• Industry Influence: With a background at Tesla and Leap Motion, Murati is widely regarded as a prominent figure in AI, known for her collaborations with Microsoft and OpenAI’s CEO, Sam Altman.

Source: https://www.reuters.com/technology/artificial-intelligence/former-openai-technology-chief-mira-murati-raise-capital-new-ai-startup-sources-2024-10-18/

How To Use AI Lip Sync on Kling AI

Follow these steps to create an AI video and sync it with audio using Kling AI:

Here's how it works:

• Visit the Kling AI website and create an account.

• Once registered, sign up to receive daily free credits, which you'll need to generate videos and use the Lip Sync feature.

• Navigate to the "AI Videos" section on the site.

• Select the option "Image to Video."

• Upload an image that fits the context of the video you want to create (e.g., if you want a woman speaking, upload an image of a woman’s face).

• In the prompt field, type something like "the woman is speaking."

• Generate the video and wait for it to process.

• Once your video is ready, click on the "Lip Sync" option.

• Upload an audio file. You can either:

  • Generate a voiceover using ElevenLabs, or

  • Use a real voice recording.

• After uploading the audio, click "Lip Sync" again. Kling AI will animate the mouth movements to perfectly match the words in the audio.

• Note that lip-syncing a 10-second video costs 10 credits.

• You can sync up to 10 seconds of video at a time, so ensure your audio fits within that limit.

That's it! You've successfully created a lip-synced AI video using Kling AI.

Top AI Tools of the Week

AI Tools for your Business -
Practical Online Program🎓

Master the most advanced AI tools to supercharge your processes, minimize costs, and explode your revenue. Whether you’re a business leader or an ambitious professional, this program will equip you with the skills and strategies to harness the full potential of AI.