#AI research

#AI research

OpenAI CEO Sam Altman was fired from his position, and Ilya Sutskever, the company's chief scientist and co-founder, is taking on a more prominent role.

The departures of Altman and co-founder Greg Brockman were reportedly due to a breakdown in communication with the board, not related to malfeasance or financial issues.

Sutskever has a background in AI research and has worked with influential figures like Geoffrey Hinton and Andrew Ng. [ more ]

ODSC curates impactful AI developments each week. [ more ]

The UK's Isambard-AI supercomputer, one of the fastest globally, will drive AI research, including in language models, healthcare, robotics, automated drug discovery, and climate research. [ more ]

Large language models (LLMs) could potentially be modified to operate in space for communication with extraterrestrial life, a lighthearted proposal by AI researcher Andrej Karpathy. [ more ]

Kylie Robison joins The Verge as senior AI reporter to lead AI coverage, working closely with policy and tech teams. [ more ]

AI investment and experimentation set some banks ahead, like JPMorgan Chase, Capital One, and Royal Bank of Canada. [ more ]

Open sourcing xAI's Grok chatbot on GitHub and Hugging Face.

Grok chatbot is notable for its vast 314 billion parameters. [ more ]

Nvidia unveils LATTE3D

LATTE3D allows near-real-time text-to-3D generation [ more ]

Generative AI tools are gaining traction in businesses.

Limited research on AI's measurable impact on team performance. [ more ]

Generative AI is used to review machine learning work of peers in academia.

Authors found an increase in large language model (LLM) usage in peer reviews submitted close to deadlines. [ more ]

AI researchers call for independent evaluation of AI models

Concerns about restrictions hindering critical research [ more ]

Google introduced new Gemma language models smaller than its Gemini Ultra model, optimized for laptops and desktop workstations.

Gemma models are accessible through Hugging Face and provide pretraining weights, inference code, and fine-tuning code. [ more ]

Users experienced unexpected responses from ChatGPT leading to confusion and concern.

OpenAI acknowledged the issue and is investigating the unexpected behavior of ChatGPT. [ more ]

California and New York are using legislation and executive action to incentivize AI research while implementing regulations to address AI harms.

New York Governor Kathy Hochul has proposed new laws to criminalize the dissemination of fabricated media, unauthorized use of someone's voice in advertising, misuse of AI, and digitally manipulated false images. [ more ]

Yann LeCun was awarded the Turing Award in 2018 for his work on deep neural networks.

LeCun and his team at Meta are now focused on building new AI products. [ more ]

The Allen Institute for AI (AI2) has released open language models (OLMo) and a large public dataset (Dolma) for text-generating AI.

The OLMo models come with code, training data, and evaluation metrics, making them more truly open than other models. [ more ]

The results of a survey suggest that AI researchers believe there is a 5% chance of AI wiping out humanity.

There are concerns that the survey results may be biased towards an alarmist perspective due to the funding sources and framing of questions. [ more ]

MIT researchers introduce AI method built from pre-trained language models

Apple reorganizes AI team to merge San Diego and Texas employees [ more ]

The National Science Foundation and other government agencies are partnering with AI developers to launch a pilot program called NAIRR.

NAIRR provides access to AI models, computing power, datasets, software, and training for US-based AI researchers. [ more ]

The US National Science Foundation has launched the National Artificial Intelligence Research Resource (NAIRR) to help academics access computing power, data, and tools for AI research.

More than 20 companies and organizations, including Google, Nvidia, OpenAI, and Amazon, have pledged to share their resources with researchers accepted into the program. [ more ]

Meta is reorganizing its AI teams to focus on creating artificial general intelligence (AGI) and improving the metaverse.

The company is merging its Fundamental AI Research (FAIR) team and its team focused on generative AI experiences. [ more ]

FAIR, Meta's AI research team, will now sit in its product organization.

Joelle Pineau and Yann LeCun will report to Meta's chief product officer, Chris Cox. [ more ]

Fingerprint ridges on different fingers of the same person can have the same angles and curvature at the center.

A deep contrastive network was used to study commonalities in fingerprints and achieved an accuracy of 77% in identifying if prints were from the same person. [ more ]

Five percent of AI experts believe AI will lead to collapse, while 10 percent believe there's a 25 percent chance it will destroy the human race.

Possible causes cited by experts include AI enabling threatening groups, authoritarian rulers using AI for control, and AI worsening economic inequality. [ more ]

A majority of AI researchers believe there is a 5% chance of human extinction or other catastrophic outcomes from AI.

Researchers anticipate AI's chances of completing certain tasks within a decade to be 50% or higher, but more complex tasks will take longer. [ more ]

Large language models (LLMs) like ChatGPT are driving innovative research and applications.

Retrieval augmented generation (RAG) enhances the accuracy of generated responses by integrating external knowledge.

The open source project Akcio utilizes the RAG approach to create a robust question-answer system. [ more ]

AI researchers at Stanford are using data from computational fluid dynamics to advance rocket science, oceanography, and climate modeling.

They aim to create a single dataset that can revolutionize these fields similar to how web-scale data revolutionized language.

The project could lead to breakthroughs in understanding complex fluid dynamics and improving predictions in various domains. [ more ]

GPT-4, a large language model by OpenAI, demonstrated emergent behavior by creating a primitive image of a unicorn when asked to draw one

Emergent behavior is the ability of AI models to exhibit unexpected abilities beyond their training

Researchers are studying how these emergent abilities occur in language models [ more ]

Tech companies are investing in AI research and integrating AI into their products

Semantic search with vector databases and democratization of AI are emerging AI tools [ more ]

The National AI Research Resource provides researchers with computing power and government datasets to advance AI.

Stanford AI leaders gathered to discuss the importance of the National AI Research Resource. [ more ]

OpenAI is aggressively recruiting senior researchers from Google's AI teams with the promise of annual pay ranging from $5 million to $10 million, mostly in stock.

OpenAI is planning to sell shares of its existing employees, which could value the company at $80 billion to $90 billion.

OpenAI primarily makes its money through subscriptions to ChatGPT Plus and by selling use of its AI models to developers and businesses. [ more ]

The Stanford Institute for Human-Centered AI funds innovative projects that leverage AI to address scientific or societal challenges.

One of the funded projects uses AI to explore the social lives of concepts to better understand culture and identify effective interventions. [ more ]

AI researchers at Stanford are collecting data from computational fluid dynamics to create a dataset for rocket science, oceanography, and climate modeling.

They hope to use this dataset to advance these scientific fields in the same way web-scale data advanced language processing.

This project demonstrates the potential of AI and data collection to revolutionize scientific research. [ more ]