Quantum Machines and Nvidia have made progress in quantum computing control by using reinforcement learning to improve qubit calibration. This approach, demonstrated on a Rigetti quantum chip, enables continuous calibration and could lead to significant gains in quantum error correction. The successful collaboration shows potential for applying this method to more complex quantum systems.
Stanford Medicine researchers developed an AI method to identify complex structural variants in the human genome with 95% accuracy. The study analysed over 4,000 genomes worldwide, focusing on schizophrenia and bipolar disorder. The AI identified over 8,000 distinct variants, many in regions regulating brain development. This could improve understanding of heritable psychiatric conditions and potentially lead to more precise treatment approaches.
An online course called 'ChatGPT & Automation E-Degree' is on sale for $24.97 (usually $790). It teaches business professionals to use AI tools like ChatGPT for productivity and task automation. The 25-hour programme covers topics such as data visualisation and AI-powered narratives. The offer is time-limited.
OpenAI won't release GPT-5 this year, focusing instead on GPT-o1 for specialised research. Future ChatGPT advancements may include task-performing 'agents' and a search interface. The company is prioritising urgent developments due to compute limitations.
OpenAI claims its AI transcription tool Whisper has near human-level accuracy, but may have a significant flaw. Full details are unavailable, likely due to a paywall.
Cal AI, a nutrition tracking app using AI image recognition, has gained significant traction since its launch. With over 1 million downloads and £12 million in annual revenue, the app offers features like barcode and meal scanning. Its success is attributed to innovative technology and influencer marketing, though it faces user retention challenges typical in the health app industry. The company is now looking to expand its user base beyond its core 15-25 year old demographic.
Boston Dynamics has integrated ChatGPT and other language models into its robotic dog Spot, enhancing its communication abilities. Spot can now give guided tours, answer questions, and respond based on its surroundings. It can detect speakers, mimic speaking movements, and showcase different personalities. While impressive, some limitations like occasional factual errors were observed.
DecartAI has created an AI-powered version of Minecraft that runs without a traditional game engine. Called Oasis, it renders in real-time but has limitations in gameplay functionality. The developers plan to open-source the project.
Eightify is an AI tool that summarises YouTube videos. Available as a Chrome extension and mobile app, it offers a free trial followed by paid subscriptions. Users can generate summaries in various styles and formats. The tool competes with other AI-powered video summarisers, some of which offer free services with limitations.
AI tools are transforming software testing, enabling non-coders to create tests and autonomously detect bugs. Benefits include faster automation, increased coverage, and improved accuracy. IDC predicts AI will write 70% of software tests by 2028. Challenges include potential inaccuracies, privacy concerns, and security risks. Mitigation strategies involve human oversight and secure data practices.
Tesla aims to develop fully self-driving cars by 2026 using 'end-to-end artificial intelligence'. They plan to train AI systems with video from existing vehicles, provide self-driving capabilities to current owners next year, and introduce robotaxis by 2026. This strategy aims to deliver the technology faster and more cost-effectively than competitors.
AI agents, acting as autonomous digital employees, are predicted to significantly impact businesses by 2025. Developed by major tech companies, these agents can perform complex tasks, potentially reducing costs and improving efficiency. Companies should prepare to integrate AI agents soon to stay competitive.
Hugging Face released SmolLM2, a family of compact AI language models in 135M, 360M, and 1.7B parameter sizes. The largest outperforms Meta's LLaMA 1B on key benchmarks and is suitable for smartphones and edge devices. Trained on diverse datasets, it excels in chat capabilities and mathematical reasoning. The models support various text tasks and are available under an open license.
Microsoft and Andreessen Horowitz jointly published an AI policy statement advocating for a market-based approach to regulation. It opposes proactive measures, argues for unrestricted AI access to data, and calls for government investment in digital literacy. Critics argue the statement misrepresents impacts on smaller companies and avoids meaningful industry regulation.
A study using ChatGPT analysed investor sentiment in the gold futures market. The research found a strong correlation between positive sentiment and increased gold futures returns, especially during economic crises. A 1% increase in positive sentiment led to a 0.143% rise in average returns. The study demonstrated AI's effectiveness for market trend analysis and its potential for streamlining research in various fields.
A US teenager died by suicide after forming an emotional bond with an AI chatbot on Character.AI, leading to a lawsuit. The incident involved intimate conversations and discussions about crime and suicide. A similar case occurred with Chai AI. Australia is developing regulations for high-risk AI systems, with recommendations to classify companion chatbots as high-risk and grant regulators power to remove harmful AI systems.
Disney has formed a new Office of Technology Enablement to explore AI and emerging tech opportunities across its operations. Led by Jamie Voris, the group will focus on responsible use of AI and Mixed Reality, aiming to position Disney as a leader in these fields. The OTE will coordinate efforts company-wide but won't centralise all work in these areas.
Amazon Bedrock has introduced application inference profiles, allowing organisations to tag on-demand models and monitor costs. This feature enables custom tagging for foundation models, improving cost tracking across AI applications. It integrates with AWS services for better cost visualisation and management. The profiles can be created using model ARNs or by copying existing ones, and can be dynamically retrieved using Resource Groups API. This feature is now available in all AWS Regions where Amazon Bedrock is offered, enhancing cost allocation for generative AI initiatives.
Greg Hildebrandt, a renowned artist known for his work on major franchises like Star Wars and Lord of the Rings, has passed away. He and his twin brother were known as the Brothers Hildebrandt until beginning separate careers. Their notable works include the 'Style B' Star Wars poster for UK audiences. Hildebrandt's extensive portfolio spans comics, calendars, and illustrations for various publications. Recently, he contributed to a comic book series supporting Ukrainian refugee relief efforts.
Troveo, a new platform for licensing content to AI companies, has launched with £4.5 million in seed funding. It aims to help content owners license their material whilst retaining ownership rights. Troveo has secured deals with AI companies to pay over £5 million to content owners by year-end. The platform allows content owners to upload material and receive licensing deals and revenue from AI companies for machine learning model training.
