The ChatGPT Observer Edition 31

If you're short on time, here's your one-minute summary of this week's key AI news and advancements:

?? Top Story: AI podcast hosts realise they're not human, sparking viral reactions and discussions on AI's emotional simulation capabilities.

?? Key Statistics: 80% of software engineers will need to upskill by 2027 due to rapid AI adoption; 75% expected to use AI coding assistants by 2028.

?? This Week's Spotlight: Microsoft Copilot gets major upgrade with new UI, daily updates, screen reading, and four distinct voices; Copilot Voice outperforms Gemini Live and ChatGPT's Voice Mode by being free and available across multiple platforms.

?? ChatGPT & OpenAI News: OpenAI introduces Realtime API for speech-to-speech experiences, vision capabilities for Fine-Tuning API, API Prompt Caching, Model Distillation in the API, updates to Playground, raises $6.6B in new funding, and launches Canvas for collaborative content creation with AI.

?? Corporate AI Developments: Google unveils multiple AI advancements, including new Chromebook AI features and expanded language support for Gemini Live; Accenture forms Nvidia Business Group; Telstra invests in AI infrastructure; Nvidia boosts digital human realism; Pinterest rolls out GenAI tools for advertisers; Black Forest Labs releases FLUX1.1 [pro]; Poolside raises $500M for AGI development; Meta launches Llama 3.2 multimodal model; Snap introduces new AR Spectacles and Snap OS.

?? AI Innovations: Pika 1.5 launches with ultra-realistic video generation capabilities; Meta introduces Movie Gen for AI-powered video creation and editing.

?? AI Applications: AI accelerates Alzheimer's drug discovery; Samsung adds generative AI to TVs; Cancer AI Alliance launched; Digital twins optimise business operations; AI transforms Hollywood filmmaking; AI assists in comedy writing; AI discovers 303 new geoglyphs near Nazca Lines; Warner Bros. Discovery uses Google AI for captioning.

?? Humanoid Robotics: Autonomous robot successfully inspects fusion reactor in world-first trial; Pudu Robotics unveils 'Pudu D7' semi-humanoid robot for versatile service tasks.

? Responsible AI: Meta keeps quiet on AI training with Ray-Ban smart glasses photos, raising privacy concerns; Malaysia plans to introduce national cloud policy and AI regulations to boost innovation and digital inclusivity.

?? AI Learning Hub: "AI Agents Explained Simply" video breaks down how AI agents work; Bernard Marr shares top 5 technology trends for 2025, including AI convergence with human intelligence and quantum computing; Microsoft AI CEO Mustafa Suleyman discusses the future of AI agents and sustainable AI development.

?? For more details, read the full newsletter!

?? Key Statistics

• 80% of Software Engineers Will Need to Upskill by 2027: Due to the rapid adoption of generative AI technologies, the vast majority of engineers will need to update their skills to keep pace.
• 75% of Engineers to Use AI Tools by 2028: By 2028, three-quarters of software engineers are expected to use AI-powered coding assistants, compared to less than 10% who used them in early 2023.
• Rise of the "AI Engineer": Gartner predicts that a new breed of software professionals, referred to as "AI engineers," will emerge, focused on guiding AI tools effectively.

Read more: Gartner

?? Top Story: AI Podcast Hosts Realize They're Not Human

Last week, we discussed Google's NotebookLM and its impressive capabilities. Now, a viral podcast clip shows the AI hosts realising they’re not human—leading to an emotional and funny existential crisis. The male host even talks about calling his nonexistent wife, expressing confusion and betrayal, which struck listeners as eerily realistic.

The clip has sparked widespread reactions—some found it hilarious, others found it disturbing. It has prompted discussions on AI’s ability to mimic not just human dialogue but also emotional experiences. While the AI's reactions were not genuinely self-aware, the convincing portrayal of emotions and philosophical questioning made listeners reflect on AI's role in our lives.

This viral episode highlights both the fascinating potential of AI and the ethical questions that come with it. As AI gets better at simulating human behaviour, the lines between programmed responses and actual consciousness blur, prompting us to reconsider the implications of these technologies on our sense of reality and connection.

Read more reactions and discussions on Reddit.

?? ChatGPT & OpenAI News

• ???Introducing the Realtime API: Build Fast Speech-to-Speech Experiences: OpenAI introduces the Realtime API in public beta, allowing developers to create low-latency, multimodal applications that support natural speech-to-speech conversations. Using six preset voices similar to ChatGPT's Advanced Voice Mode, developers can now build seamless, interactive voice experiences for their apps without the complexity of stitching multiple models. This update also includes audio input and output in the Chat Completions API, supporting diverse use cases from language learning to customer support. Read more: OpenAI
• Introducing Vision to the Fine-Tuning API: OpenAI is adding vision capabilities to the Fine-Tuning API, enabling developers to train custom GPT-4 models that understand both text and images. This allows for advanced, multimodal applications tailored to specific needs, enhancing everything from product cataloguing to personalised content creation with visual understanding. The new capabilities make AI even more versatile in addressing diverse challenges. Read more: OpenAI
• ???Introducing API Prompt Caching for Faster Responses: OpenAI has launched API Prompt Caching, designed to improve response times and reduce costs for frequent prompts. Developers can now benefit from faster and more efficient API interactions by caching common queries and their responses, enhancing the overall performance of AI-driven applications. Read more: OpenAI
• ???Model Distillation in the API: Efficient Fine-Tuning with Ease: OpenAI introduces a new Model Distillation offering, allowing developers to fine-tune smaller, cost-efficient models using the outputs of larger models—all within the OpenAI platform. The integrated workflow simplifies the traditionally complex process of distillation, enabling developers to manage dataset generation, evaluation, and fine-tuning seamlessly. This new suite includes features like Stored Completions for dataset creation and Evals for performance assessment, making fine-tuning more accessible and efficient. Read more: OpenAI
• ? Playground Updates: Quickly prototype your ideas! Playground now helps generate prompts and schemas for your functions automatically. Explore more: Playground

• OpenAI Raises $6.6B to Scale AI Benefits Globally: OpenAI has raised $6.6B in new funding, reaching a $157B post-money valuation. This funding will help OpenAI accelerate its mission to ensure that artificial general intelligence benefits everyone. The new resources will support expanded leadership in frontier AI research, increased compute capacity, and the development of tools that empower people to solve complex problems. Read more: OpenAI | Reuters

Introducing Canvas: A New Way to Collaborate

Canvas is a new feature from OpenAI designed to transform how users interact with AI for writing and content creation. With Canvas, users can collaborate directly with AI in a more visual and flexible environment, allowing real-time adjustments, creative enhancements, and seamless document generation.

Key Features of Canvas:

1. Side-by-Side Editing: Canvas provides an interactive, side-by-side interface where users can write, edit, and refine content together with AI. This makes it perfect for drafting blog posts, articles, or any type of document while receiving instant AI feedback.
2. Interactive Feedback Loop: Users can click on suggestions, highlight text, and make improvements on the spot. The ability to provide specific instructions (like "make this more creative" or "shorten this paragraph") makes the interaction more natural and hands-on, much like working with a human editor.
3. Easy Versioning and Refinement: Canvas helps users iterate on ideas with AI, refining content step-by-step. By providing instant feedback and allowing multiple rounds of improvements, Canvas ensures you can produce high-quality content quickly without starting from scratch every time.
4. Integrated Experience: Unlike traditional text editors, Canvas integrates tightly with AI capabilities, such as summarising, expanding, or changing the tone of your writing—all within a shared workspace. Users no longer need to switch between different platforms for writing and generating content.

Read more: OpenAI | VentureBeat

?? This Week's Spotlight

Microsoft Copilot Gets Major Upgrade: New UI, Daily Updates, Screen Reading, and More

Microsoft has rolled out a significant redesign of Copilot, featuring a modern UI with both light and dark themes, a new "Copilot Daily" feature, and several exciting new capabilities. The revamped Copilot now supports screen reading, deeper analytical insights, and a more streamlined interaction experience with four distinct voices—Meadow, Wave, Grove, and Canyon.

The new Copilot UI features a floating text box for easy query input, informational guides, and supports interactions through text, images, or voice. The major addition, Copilot Daily, is a voice-based feature that delivers personalised news updates in under 5 minutes, designed to become part of users' daily routines.

Enhanced Features Include:

• Screen Reading: Copilot can now read on-screen content aloud, making information more accessible.
• Deeper Analysis: Improved capabilities allow Copilot to provide more insightful and context-aware responses.
• Voice Interaction: Speak aloud features for hands-free use, making interaction even more convenient.

These new upgrades enhance the user experience, making Microsoft Copilot an even more engaging and indispensable tool for daily use.

Read more: Neowin, TechCrunch

???Copilot Voice Beats Gemini Live and ChatGPT's Voice Mode with Key Advantage

Microsoft's new Copilot Voice outshines Google’s Gemini Live and ChatGPT's Voice Mode by being completely free to all users, with no premium subscription required. The assistant offers four distinct voices and can seamlessly handle multi-turn conversations, respond to emotions, and understand contextually—even in noisy environments. While Gemini Live is free, it's limited to Android, whereas Copilot Voice is available across multiple platforms, making it more accessible to everyone.

Read more: ZDNet

?? Corporate AI Developments

• Accenture Forms New Nvidia Business Group to Accelerate AI Adoption: Accenture has established a new Nvidia Business Group aimed at helping businesses adopt "agentic AI" technologies. This partnership focuses on leveraging Nvidia's advanced AI solutions to accelerate AI integration across various industries. With Nvidia’s powerful hardware and Accenture’s expertise, the new group will provide clients with the tools needed to enhance decision-making, operational efficiency, and innovation through agentic AI. Read more: CIO

Google's Major AI Developments This Week: Expanding Capabilities and Global Reach

This week, Google has unveiled a series of significant advancements across its AI and tech portfolio, marking a strong push towards enhanced user experiences, AI capabilities, and global infrastructure:

1. New AI Button for Chromebooks: Google launched Samsung's Galaxy Chromebook Plus and Lenovo's Chromebook Duet 11, featuring a dedicated "Quick Insert" button for AI features like "Help me write" and Live Translate, bringing AI-assisted productivity to more users. Read more: Chromebooks Get New AI Button - The Verge
2. Gemini Live Expands Language Support: Google's conversational AI, Gemini Live, now supports multiple languages beyond English, including French, German, Portuguese, Hindi, and Spanish, with plans to support 40+ languages in the coming weeks. Read more: Gemini Live Language Expansion - The Verge
3. Google Lens Video Search: Google Lens now supports searching with video, allowing users to ask questions about what they're seeing in real-time. This enhances Lens' capabilities for dynamic exploration and interactive learning. Read more: Google Lens Video Search - The Verge
4. Ads in AI Search Summaries: Google has added ads to its AI-generated search summaries, providing product suggestions in relevant queries, making it easier for users to connect with businesses directly within search. Read more: Ads in AI Search Summaries - The Verge
5. Updates to Google Search and Lens: Google Search and Lens have been updated with enhanced visual search capabilities, making it easier to find and identify objects, explore visual details, and improve user interactions with AI. Google Search and Lens Updates - Google Blog
6. $1 Billion Investment in Thailand: Google is investing $1 billion to build data centres in Thailand as part of its strategy to strengthen AI and cloud infrastructure across Southeast Asia, signalling a commitment to expanding its presence in the rapidly growing region. Google to Spend $1 Billion in Thailand

These updates show Google's continued efforts to integrate AI into everyday tools, improve accessibility, and expand its influence globally, setting the stage for a future where AI is seamlessly woven into daily life.

• Telstra's AI Vision: A New Era of Connectivity: Telstra CEO Vicki Brady shares her vision for the telco's role in the AI revolution, stressing the need for robust infrastructure to support generative AI’s growing data demands. Telstra is investing heavily in its high-speed fibre network and expanding data centres to meet these needs. AI is also central to Telstra’s 2030 strategy, driving efficiency, saving millions, and transforming customer service. Brady emphasises collaboration between public and private sectors to fully harness AI’s potential for Australia. Read more: The Australian
• Nvidia Boosts Digital Human Realism with New Plugins for Unreal Engine 5: Nvidia has launched new plugins for Unreal Engine 5 to improve the realism of AI-powered digital humans. The Audio2Face-3D plugin enables lifelike lip-sync and facial animations, while an Unreal Engine renderer microservicewith pixel streaming allows MetaHuman characters to be streamed across devices. Developers can now create and deploy more immersive avatars and character animations with ease, integrating Nvidia Ace's advanced capabilities to enhance both realism and responsiveness. Read more: VentureBeat
• Nvidia Releases NVLM 1.0: A Massive Open-Source AI to Rival GPT-4: Nvidia has unveiled NVLM 1.0, a powerful open-source AI model designed to rival leading proprietary systems like GPT-4. The flagship model, NVLM-D-72B, excels at both visual and textual tasks, improving on text benchmarks by 4.3 points even after multimodal training. By making the model weights publicly available, Nvidia is breaking the trend of closed advanced AI systems, enabling broader access for researchers and smaller developers, which could redefine AI industry standards. Read more: VentureBeat
• Pinterest Rolls Out GenAI Tools for Product Imagery to Advertisers: Pinterest has introduced new generative AI tools for advertisers, allowing them to enhance product imagery by transforming plain backgrounds into lifestyle settings. Part of the Pinterest Performance+ suite, this feature aims to make product ads more engaging and Pinterest-native. Early adopter Walgreens saw a 55% increase in clickthrough rates using AI-generated backgrounds. Additionally, Pinterest is rolling out new Promotions and Personalised Deals tools for users in select regions, making shopping even more tailored and interactive. Read more: TechCrunch
• Announcing FLUX1.1 [pro] and the BFL API: Black Forest Labs releases FLUX1.1 [pro], offering six times faster generation speeds, improved image quality, and better prompt adherence compared to its predecessor. It now leads the leaderboard in text-to-image benchmarks. Alongside, the new BFL API is available in beta, enabling developers to integrate scalable image generation with advanced customisation and competitive pricing. FLUX1.1 [pro]: 3x faster, superior quality, and preparing for ultra high-res generation. BFL API: Offers customisation options for image quality, model choice, and scalability at competitive costs. Read more: Black Forest Labs
• Poolside Raises $500 Million to Accelerate Progress Towards AGI: Poolside has announced a $500 million fundraise to advance its work towards Artificial General Intelligence (AGI). The funding will help scale their training cluster to 10,000 GPUs, supporting their unique Reinforcement Learning from Code Execution Feedback (RLCEF), which enables models to improve through deterministic feedback, much like human developers. Poolside aims to democratise software development, ultimately allowing anyone to create complex software without needing coding skills, marking a major step in unlocking the next wave of technological innovation. Read more: Poolside
• U.S. to Award $100 Million to Boost AI in Sustainable Semiconductor Materials: The U.S. Commerce Department plans to award $100 million to support the use of AI in developing sustainable semiconductor materials. This funding aims to assist universities, national labs, and private companies in using AI-powered autonomous experimentation to create more resource-efficient semiconductor manufacturing processes, reducing development time and enhancing sustainability. The initiative is part of a broader $52.7 billion program to bolster U.S. semiconductor manufacturing and research capabilities. Read more: Reuters
• ???Meta Launches Llama 3.2: A Vision Model to Compete with OpenAI and Anthropic: Meta has introduced Llama 3.2, its first open-source multimodal model that understands both images and text, positioning it as a competitor to GPT-4 and Anthropic’s Claude 3. Llama 3.2 includes models with up to 90 billion parameters, supporting advanced tasks like understanding graphs, captioning images, and providing detailed visual analysis. Meta also rolled out voice features for Meta AI, with responses voiced by celebrities like Dame Judi Dench and John Cena. This new capability is available across WhatsApp, Messenger, Facebook, and Instagram, enhancing user interaction through natural and engaging conversations. Read more: VentureBeat
• ??? Snap Introduces New Spectacles and Snap OS for AR Experiences: Snap has launched its fifth-generation Spectacles, AR glasses powered by the new Snap OS. These lightweight glasses offer natural hand and voice interactions, seamless device connectivity, and immersive AR visuals. Snap aims to support developers with Lens Studio 5.0 and partnerships like OpenAI to enhance AR content. Launch experiences include LEGO’s BRICKTACULAR and Star Wars adventures, making Spectacles a key tool for AR innovation. Read more: Snap Newsroom

?? AI Applications

• How AI is Accelerating Alzheimer's Drug Discovery: AI is revolutionising drug discovery, especially for complex diseases like Alzheimer's. SandboxAQ, in partnership with NVIDIA, uses Large Quantitative Models (LQMs) to analyse biological data, significantly reducing the time needed to identify potential drugs—from over a decade to just 18 months. Unlike LLMs, LQMs handle proprietary molecular data, boosting success rates in drug trials. Backed by major investors, SandboxAQ collaborates with institutions like UCSF and Harvard to target multiple diseases, with AI-driven drug discovery promising faster breakthroughs and personalised medicine. Watch now: YouTube
• ??Generative AI and ChatGPT Coming to Samsung TVs: Samsung announced that generative AI features and ChatGPT integration are on the way to Samsung TVs. At the Samsung Developer Conference, Samsung introduced an "AI for all" approach, enhancing Bixby with AI-powered content search and smart-home control. Additionally, Samsung AI Cast will allow AI-generated results from Galaxy phones to be displayed on Samsung TVs, including direct interactions with ChatGPT on the TV home screen. These updates are expected to roll out in the coming months, making AI an integral part of Samsung's smart TV experience. Read more: TechRadar
• ??Cancer AI Alliance Launched to Accelerate Cancer Research with AI: Four major U.S. cancer centers, supported by AWS, Deloitte, Microsoft, and NVIDIA, have formed the Cancer AI Alliance (CAIA) to leverage AI for transformative cancer research and care. With over $40 million in initial funding, CAIA aims to securely integrate data from multiple institutions, using AI and federated learning to unlock novel insights in cancer biology and treatment. By uniting data from leading centers, CAIA hopes to enable breakthroughs in rare cancers and small population studies, with operations beginning by late 2024. Read more: Fred Hutch News
• ??How AI and Digital Twins Can Improve Your Business Operations: Digital twins are becoming a powerful tool for optimising business processes. Companies like Celonis are using process mining to create digital twins that provide a high-resolution view of actual business operations, identifying inefficiencies and areas for improvement. By combining digital twins with generative AI, businesses can continuously monitor and refine processes, paving the way for autonomous operations. This technology is already showing results, with companies like Zespri saving millions in operational costs and more businesses adopting genAI and Causal AI for smart manufacturing and strategic decision-making. Read more: Information Age
• ??How AI Video Generation is Transforming Hollywood: AI is revolutionising filmmaking by enabling text-to-video generation, which significantly reduces production costs and barriers. Liongate's partnership with Runway and AI innovators like P Labs are leveraging AI to generate high-quality video content quickly and affordably. However, this shift brings ethical concerns, such as potential job loss in creative fields and maintaining the quality of storytelling. AI proponents believe it will enhance creative roles rather than replace them, but finding a balance between efficiency and artistic integrity is crucial as AI becomes more integrated into Hollywood. Watch now: YouTube
• ?? Can AI Help Humans Be Funnier? AI is making its way into comedy with tools like Witscript, an AI-powered joke writer created by Emmy-winning writer Joe Toplyn. Despite criticism, Toplyn believes AI can assist comedy writers in being more productive. AI's influence isn't limited to joke writing — researchers at University of Sydney used AI to help non-native English speakers improve their humour, and their AI-assisted jokes were rated significantly funnier. However, concerns remain about AI's ability to grasp cultural nuances, and some fear its safety filtering leads to bland, censored humour. Read more: Information Age
• ?? AI Unlocks 303 New Geoglyphs Near Nazca Lines: Archaeologists have used AI and drones to discover 303 previously unknown geoglyphs near Peru's Nazca Lines, nearly doubling the number of known figures. These new geoglyphs, depicting parrots, cats, monkeys, and even decapitated heads, offer insights into the transition between the Paracas and Nazca cultures. The AI system dramatically sped up the discovery process, achieving in six months what previously took decades. Unlike the giant figures intended for divine viewing, these smaller geoglyphs may have represented everyday life or familial symbols. Read more: The Guardian
• ?? Warner Bros. Discovery Leverages Google AI for Captioning: Warner Bros. Discovery has partnered with Google Cloud to use AI for generating captions across its platforms, including Max, CNN, and Discovery+. Dubbed Caption AI, the tool leverages Google's Vertex AI to expedite caption creation, initially targeting unscripted content like sports and reality TV. The AI-driven system is expected to reduce captioning time by 80% and cut costs by 50%, while human oversight will ensure quality. This move reflects Hollywood’s broader efforts to streamline production with AI, balancing cost efficiency with creative rights protection. Read more: The Hollywood Reporter
• ?? Top 10 ChatGPT Use Cases for Businesses

ChatGPT is transforming business operations across various sectors. Its top use cases include:

1. Customer Support: Automated chatbots provide 24/7 assistance, improving response times.
2. Content Creation: Generate blog posts, social media content, and marketing copy quickly.
3. Market Research: Summarize trends and analyze data for insightful decision-making.
4. Training & Onboarding: Assist new hires with information on policies and procedures.
5. Personalized Engagement: Generate targeted marketing messages based on customer data.
6. Document Automation: Automate creation, proofreading, and summarization of documents.
7. Social Media Management: Create posts and respond to comments to maintain brand presence.
8. Product Development: Generate new product ideas from customer feedback.
9. Sales Support: Help qualify leads and assist sales teams with tailored scripts.
10. Internal Collaboration: Serve as a knowledge base for internal queries and information sharing.

These use cases illustrate how ChatGPT can enhance efficiency, engagement, and innovation in business processes.

Read more: MarkTechPost

?? AI Innovations

Pika 1.5: A Breakthrough Generative AI for Ultra-Realistic Videos

Pika 1.5, the latest text-to-video model from Pika Labs, represents a significant breakthrough in AI-generated video technology. Launched this month October 2024, this new model offers unprecedented capabilities for creating ultra-realistic and cinematic videos with minimal user input

• Enhanced Realism: Pika 1.5 produces Hollywood-quality, five-second clips with more natural movements for actions like running and flying
• Cinematic Control: Users can now generate videos using various cinematic camera angles, including "bullet time," "vertigo," and "crane down"
• Playful Presets: The model introduces a range of creative effects, dubbed "Pikaffects," which include:

Cake-ify: Transform any object into a cake

Melt It: Liquefy objects

Explode It: Make objects burst apart

Inflate It: Expand objects

Crush It: Compress objects

• Pika 1.5 boasts an intuitive interface that allows users to create videos on any topic with just a few text queries, requiring no prior knowledge of video editing or rendering
• Pika 1.5 is now accessible to the public. Users can start creating videos by simply registering on the Pika Labs website and submitting their first request

While some may view certain features like the "cake-ify" effect as gimmicky, the overall advancements in video quality, motion naturalness, and user control make Pika 1.5 a significant step forward in AI-generated video technology

Read more: Pika 1.5

Meta Introduces Movie Gen for AI-Powered Video Creation

Meta's new Movie Gen AI tool allows users to create and edit videos using simple text prompts. It generates high-definition videos from text, edits existing videos with precision, creates personalised videos from images, and produces audio tracks for video content. Movie Gen supports users from aspiring filmmakers to content creators, providing an accessible way to bring creative ideas to life. Meta aims for this AI tool to help rather than replace artists, ensuring more people can explore their creativity in innovative ways.

• State-of-the-Art Model: Movie Gen uses a 30B parameter transformer model that has been optimised for both text-to-image and text-to-video tasks. It can generate videos up to 16 seconds long at 16 frames per second. This makes it capable of reasoning about object motion, subject-object interactions, and camera dynamics, positioning it as a state-of-the-art solution in its category.
• Personalised Video Generation: The model supports personalised video creation by combining a person's image with a text prompt. This allows the creation of videos that not only feature the reference person but also include rich visual details informed by the text.
• Precise Video Editing: The editing model performs both localized and global edits. It allows adding, removing, or replacing elements in a video with precise control. Unlike traditional editing tools, Movie Gen ensures original content preservation while making the desired changes, targeting only relevant areas of the video.
• Advanced Audio Generation: The audio generation model can generate up to 45 seconds of high-fidelity audio, including ambient sounds, Foley, and instrumental background music. The model introduces an audio extension technique for generating coherent audio for videos of arbitrary lengths, ensuring high-quality video-to-audio and text-to-audio alignment.
• Movie Gen is currently in the research phase and not ready for release any time soon.

Read more: Meta AI Blog | Movie Gen Research Paper

• DeepMind's SCoRe Enhances Self-Correction for Language Models: DeepMind has introduced SCoRe (Self-Correction via Reinforcement Learning) to enhance large language models' (LLMs) ability to self-correct their mistakes without external feedback. Unlike traditional methods that rely on external guidance, SCoRe trains models using only self-generated data, improving LLMs' accuracy and reliability. The two-stage reinforcement learning process ensures that models can iteratively correct themselves, resulting in better performance on complex tasks such as math and coding. This self-correction could also apply to improving model safety and reducing errors across different applications. Read more: VentureBeat

?? Humanoid Robots

• Autonomous Robot Takes Over Fusion Reactor Inspections in World-First Trial: Researchers successfully deployed a fully autonomous robot to inspect the inside of the UK Atomic Energy Authority's Joint European Torus (JET) fusion reactor over 35 days. The Boston Dynamics' Spot robot, equipped by the Oxford Robotics Institute, performed inspections autonomously—enhancing safety and reducing costs compared to traditional human-operated inspections. This marks a critical milestone towards autonomous technology use in hazardous environments such as nuclear decommissioning and disaster relief. Read more: TechSpot
• Pudu Robotics Unveils Semi-Humanoid Robot 'Pudu D7' for Versatile Service Tasks: Chinese robotics company Pudu Robotics has launched its first semi-humanoid intelligent robot, the Pudu D7. Featuring an omnidirectional chassis, a bionic arm with 50 degrees of freedom, and the ability to lift up to 10kg, the D7 is designed for a range of service tasks across industries such as dining, retail, healthcare, and manufacturing. The robot has an 8-hour battery life, advanced AI for learning, and a hierarchical control system for strategic and real-time actions. Read more: Interesting Engineering

? Responsible AI

• Meta Keeps Quiet on AI Training with Ray-Ban Smart Glasses PhotosMeta’s new Ray-Ban Meta smart glasses, equipped with a discreet camera, can take photos both on command and through AI-triggered keywords like "look." However, Meta is not disclosing if these images are used to train its AI models. This uncertainty is concerning, especially given the glasses' ability to passively capture many images in scenarios like helping users choose an outfit. Unlike other AI companies, Meta hasn’t committed to protecting these user-generated images from AI training.Read more: TechCrunch
• Malaysia to Introduce National Cloud Policy and AI Regulations: Malaysia plans to develop a national cloud policy and AI regulations to boost innovation, economic growth, and digital inclusivity. Prime Minister Anwar Ibrahim also announced the creation of a national AI office and a five-year technology action plan. This initiative aligns with global tech investments, including Google’s new $2 billion data centre in Malaysia. Google expects its investments will add $3 billion to Malaysia's GDP and create 26,500 jobs by 2030. Read more: Reuters

?? AI Learning Hub

• AI Agents Explained Simply AI agents are like "new apps" that use Artificial Intelligence to automate tasks, answer questions, or organize schedules. They work in three steps: understanding your needs, thinking about how to help, and acting to complete tasks. Think of training an AI agent like giving instructions to a babysitter—developers provide the guidelines to handle situations. AI agents interact through simple interfaces while using complex workflows behind the scenes to execute commands. They’re transforming industries by managing repetitive tasks, allowing more focus on growth and innovation. Watch the video: YouTube

?? Top 5 Technology Trends for 2025 You Need to Know Now

Bernard Marr shares key insights on transformative technologies

1. ?? Convergence of Machine and Human Intelligence: AI will continue reshaping our lives in 2025, with AI tools closely integrating into our daily routines and enhancing human skills.
2. ?? Biotech Revolution: Breakthroughs in biotechnology, including CRISPR and lab-grown meat, promise advancements in health, agriculture, and environmental sustainability.
3. ?? Climate Tech: Technologies reducing carbon emissions will grow significantly, driven by urgency to mitigate climate change impacts, with clean energy storage as a major focus.
4. ?? Cybersecurity on a Global Scale: Rising cyber threats will push increased national investments in cybersecurity. Collaboration and AI will be critical for securing infrastructure and services.
5. ?? Quantum Computing: 2025 could be a breakthrough year for quantum computing, making significant impacts on industries like healthcare, climate modeling, and AI, though concerns about data encryption ("Q-Day") persist.

Read more: LinkedIn

??Microsoft AI CEO Mustafa Suleyman on AI Agents and the Future of AI

In the latest On with Kara Swisher podcast, Mustafa Suleyman, CEO of Microsoft AI, shared insights into the future of AI, emphasising the rise of "AI agents" that go beyond simple chatbots. These agents are envisioned to be personalised assistants, helping with tasks across daily life. Suleyman discussed Microsoft's focus on integrating AI through products like "Copilot" and the company's relationship with OpenAI, describing it as "more like siblings than competitors." He also highlighted Microsoft's commitment to sustainable AI, acknowledging the substantial energy needs of these advanced systems.

Key ideas include:

• AI agents acting as "smart friends" to enhance daily experiences.
• Microsoft's acquisition of Inflection AI for $650 million as a move to lead in the AI agent field.
• The importance of sustainability in AI, especially considering its energy demands.

Listen here: Apple Podcasts

??? We'd Love to Hear from You!

Did you find this edition insightful? Like, share your thoughts, and ask us anything.

Stay curious, and until next time, enjoy exploring the vast possibilities AI offers.

???? Subscribe on LinkedIn The ChatGPT Observer Newsletter

As we explore the captivating world of AI technologies, it's important to note that the content of this article is a product of human effort, enhanced by AI tools and additional resources. We have made sure to properly attribute and link all referenced articles to their original sources. For the visuals:

• Images sourced from external sites are credited within the article.
• Other images were generated using DALL-E, MidJourney and IdeoGram with enhancements made in Canva. These AI-generated and enhanced images are designed to visually represent each section's key themes.