The latest in AI from November 30 to December 6, 2023

Published
Dec 6, 2023
Reading time
4 min read
AI-generated illustration of a futuristic bridge in a main city

This week's top AI news and research stories featured Amazon's new AI-powered assistant, biases in pedestrian detection models, regulations in the insurance industry’s use of AI, and a robot that helps you find your personal objects. But first:

Microsoft announces £2.5 billion investment to boost the UK’s AI capabilities
The investment aims to double Microsoft’s UK datacenter footprint by 2026, train or retrain over one million people for the AI economy, and extend Microsoft’s Accelerating Foundation Models Research (AFMR) program to prioritize GPU access for the UK’s research community. (Read more at Microsoft

Research finds opportunities and risks as heritage organizations embrace AI
A new study focuses on what innovation in AI looks like in the UK heritage sector, and showcases its diverse uses in museums, galleries, libraries, and archives. Notable examples include predictive analytics for exhibition popularity at the National Gallery. However, the study also highlighted risks such as discrimination, misinformation, copyright infringement, and transparency issues. (Read more at Museum Association)

U.S. mandates Saudi venture capital firm must sell stake in Silicon Valley AI firm 
The Biden administration has instructed Prosperity7 to sell its shares in Rain AI, a Silicon Valley AI chip startup backed by OpenAI co-founder Sam Altman. Rain AI, which designs AI chips inspired by brain functionality, had Prosperity7 as a lead investor in a funding round that raised $25 million in 2022. (Read the news story at Bloomberg

Generative AI regulation allegedly stalls EU legislation talks
Sources revealed that negotiations on foundation models have become the primary hurdle, with a risk of shelving the act before European parliamentary elections next year unless an agreement is reached. France, Germany, and Italy form an important bloc of countries opposing foundation models. Pending issues also include establishing a definition of AI and national security exceptions. Critics argue that self-regulation may fall short of safety standards for foundation models, creating legal uncertainty and impeding European industries' planning. (Read the article at Reuters)

AI fuels innovations in Pennsylvania's infrastructure projects
In Pennsylvania, U.S., where 13 percent of bridges face structural deficiencies, engineers are leveraging AI to address challenges like the development of lighter concrete blocks for construction and noise-absorbing walls along highways. The projects aim to create more resilient structures at a reduced cost. The use of AI in civil engineering could revolutionize project development, early damage detection, and real-time incident analysis, but careful consideration and regulations are urged to ensure safety and reliability. (Read the article in The New York Times)

Anduril's Roadrunner: AI combat drone takes flight 
Anduril's latest innovation combines AI technology and jet-powered capabilities to counter the escalating threat of low-cost, sophisticated aerial attacks. The modular and autonomous Roadrunner drone aims to provide rapid response and heightened resilience against evolving threats such as suicide drones. (Read more at Wired)

General Motors to reduce investment in Cruise self-driving division next year
Following recent accidents involving its self-driving taxis in San Francisco, the company, initially planning expansion to multiple cities, now focuses on rebuilding trust with regulators and communities. The decision to reduce spending follows the suspension of Cruise's robotaxi license in California and a need to regain public trust in the wake of safety incidents, including a pedestrian fatality. (Read the article at The New York Times)

Sam Altman returns as OpenAI CEO
Besides Altman’s return, Mira Murati reassumed her role as CTO, and Greg Brockman returned as President. For now, the new board comprises former Salesforce CEO Bret Taylor (Chair), economist Larry Summers, and Quora CEO Adam D’Angelo. (Read the blog post at OpenAI)

Consortium of major companies develops data provenance standards to enhance trust in AI
Many companies (including American Express, IBM, and Walmart) formed the Data & Trust Alliance, introducing new standards for data provenance in AI applications. These standards cover eight basic criteria, including lineage, source, legal rights, and data type. The goal is to offer clear data documentation and bolster efficiency and trust in AI developments. (Read more at The New York Times)

Amazon Web Services (AWS) introduces Titan models in Amazon Bedrock
Amazon’s Titan Image Generator and Titan Multimodal Embeddings offer image, multimodal, and text options through a fully managed API. The Titan Image Generator enables content creators to generate images using natural language prompts, targeting applications in advertising, e-commerce, and media. The Titan Multimodal Embeddings facilitate the creation of contextually relevant multimodal search and recommendation experiences. (Read the blog post at AWS)

Voicemod launches feature to craft and share custom synthetic voices
The app, known for its AI voice-changing program popular in the gaming and streaming communities, now enables users to craft and share their unique AI voices by modifying their own voices or choosing from various genders, ages, and tones. (Read more at The Verge)

Demand keeps soaring for prompt engineers
Prompt engineering emerged as a lucrative and sought-after skill in the year since the public launch of ChatGPT. Google searches for "prompt engineering" have skyrocketed, and LinkedIn reports substantial increases in related terms on member profiles. The skillset, involving coaxing AI systems for better results and training colleagues in using generative AI, is in high demand. Newly-created roles offer significant compensation, often upwards of $335,000 annually. (Read the analysis at Bloomberg)

Research: Deep learning model offers precision in predicting breast cancer outcomes 
The Histomic Prognostic Signature (HiPS), which evaluates both cancerous and non-cancerous cell patterns, outperformed expert pathologists in predicting disease progression. By identifying breast cancer patients classified as high or intermediate risk who could become long-term survivors, the tool offers the potential to reduce the duration or intensity of chemotherapy, sparing patients from harmful side effects. (Read the article via Northwestern University)

IBM expands geospatial AI collaboration to tackle climate challenges globally
The initiative involves mapping urban heat islands in the UAE, supporting Kenya's reforestation campaign, and enhancing climate resiliency in the UK's aviation sector. Additionally, IBM is collaborating with NASA to develop a new model for weather and climate, aiming to improve the precision and efficiency of weather forecasting and address climate-related challenges on a global scale. (Read more at IBM)

Share

Subscribe to The Batch

Stay updated with weekly AI News and Insights delivered to your inbox