tractable
RL in Latent MDPs is Tractable: Online Guarantees via Off-Policy Evaluation
In many real-world decision problems there is partially observed, hidden or latent information that remains fixed throughout an interaction. Such decision problems can be modeled as Latent Markov Decision Processes (LMDPs), where a latent variable is selected at the beginning of an interaction and is not disclosed to the agent initially. In last decade, there has been significant progress in designing learning algorithms for solving LMDPs under different structural assumptions. However, for general LMDPs, there is no known learning algorithm that provably matches the existing lower bound. We effectively resolve this open question, introducing the first sample-efficient algorithm for LMDPs without any additional structural assumptions.
Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback
Casper, Stephen, Davies, Xander, Shi, Claudia, Gilbert, Thomas Krendl, Scheurer, Jérémy, Rando, Javier, Freedman, Rachel, Korbak, Tomasz, Lindner, David, Freire, Pedro, Wang, Tony, Marks, Samuel, Segerie, Charbel-Raphaël, Carroll, Micah, Peng, Andi, Christoffersen, Phillip, Damani, Mehul, Slocum, Stewart, Anwar, Usman, Siththaranjan, Anand, Nadeau, Max, Michaud, Eric J., Pfau, Jacob, Krasheninnikov, Dmitrii, Chen, Xin, Langosco, Lauro, Hase, Peter, Bıyık, Erdem, Dragan, Anca, Krueger, David, Sadigh, Dorsa, Hadfield-Menell, Dylan
Reinforcement learning from human feedback (RLHF) is a technique for training AI systems to align with human goals. RLHF has emerged as the central method used to finetune state-of-the-art large language models (LLMs). Despite this popularity, there has been relatively little public work systematizing its flaws. In this paper, we (1) survey open problems and fundamental limitations of RLHF and related methods; (2) overview techniques to understand, improve, and complement RLHF in practice; and (3) propose auditing and disclosure standards to improve societal oversight of RLHF systems. Our work emphasizes the limitations of RLHF and highlights the importance of a multi-faceted approach to the development of safer AI systems.
- Europe > United Kingdom > England > Cambridgeshire > Cambridge (0.14)
- Europe > United Kingdom > England > Oxfordshire > Oxford (0.14)
- North America > United States > California > San Francisco County > San Francisco (0.14)
- (9 more...)
Staff MLOps Engineer at Tractable - London, UK
Tractable is an Artificial Intelligence company bringing the speed and insight of Applied AI to visual assessment. Trained on millions of data points, our AI-powered solutions connect everyone involved in insurance, repairs, and sales of homes and cars – helping people work faster and smarter, while reducing friction and waste. Founded in 2014, Tractable is now the AI tool of choice for world-leading insurance and automotive companies. Our solutions unlock the potential of Applied AI to transform the whole recovery ecosystem, from assessing damage and accelerating claims and repairs to recycling parts. They help make response to recovery up to ten times faster – even after full-scale disasters like floods and hurricanes.
Engineering Manager, MLOps at Tractable - London, UK
Tractable is an Artificial Intelligence company bringing the speed and insight of Applied AI to visual assessment. Trained on millions of data points, our AI-powered solutions connect everyone involved in insurance, repairs, and sales of homes and cars – helping people work faster and smarter, while reducing friction and waste. Founded in 2014, Tractable is now the AI tool of choice for world-leading insurance and automotive companies. Our solutions unlock the potential of Applied AI to transform the whole recovery ecosystem, from assessing damage and accelerating claims and repairs to recycling parts. They help make response to recovery up to ten times faster – even after full-scale disasters like floods and hurricanes.
Senior MLOps Engineer at Tractable - London, UK
Tractable is an Artificial Intelligence company bringing the speed and insight of Applied AI to visual assessment. Trained on millions of data points, our AI-powered solutions connect everyone involved in insurance, repairs, and sales of homes and cars – helping people work faster and smarter, while reducing friction and waste. Founded in 2014, Tractable is now the AI tool of choice for world-leading insurance and automotive companies. Our solutions unlock the potential of Applied AI to transform the whole recovery ecosystem, from assessing damage and accelerating claims and repairs to recycling parts. They help make response to recovery up to ten times faster – even after full-scale disasters like floods and hurricanes.
CIECA Forms Artificial Intelligence Committee
"AI adoption by the insurance industry is still in the infancy stage, and there is skepticism and perception about the technology," said Pofale. "The utilization of AI will undoubtedly bring huge efficiencies and cost savings and reduce cycle times. Still, the right level of education and expectation management must be done. With the rapid adoption of AI in the collision industry, it is the right time to have a common framework and best practices that will make the integrations across collision industry platforms easier." Spears is head of automotive at Tractable, an applied AI company that uses the speed and accuracy of AI to visually assess cars and homes.
Black Widow to harness Tractable's AI
InsurTech Tractable will collaborate with Black Widow, a vehicle imaging capture platform, to transform automotive sales with more transparent insights into a vehicle's condition. Black Widow uses an eight-camera, drive-through system that captures high-definition vehicle images within seconds and can be automatically edited and published online. InsurTech Tractable launched its AI tool last year, the solution can assess a vehicle's external condition and possible damage within minutes. The partnership will see Tractable's AI use the images captured by Black Widow's proprietary system to detect external damage and generate a vehicle condition report, which outlines estimated repair costs. Tractable said the combination of the companies technologies will mean auctions, dealerships and repair facilities will receive a more accurate and consistent assessment of vehicles, allowing them to make faster and better informed sales decisions.
AI opens the door to faster claims for home insurance - TechHQ
Insurance technology brings in more innovative solutions for homeowners and insurers thanks to artificial intelligence (AI). AI is disrupting the industry by allowing for faster and more user-friendly claims and creating more transparent and customized policies that suit the client's situation. This replaces the traditionally long and painful process of getting a claim or settling for a one-size-fits-all property insurance policy. AI Property from Tractable, for example, allows anyone with a smartphone to access damage quickly and efficiently to buildings caused by hurricanes, floods, and other natural disasters. Use its mobile-friendly web-based app to take photos of the external conditions and submit them to Tractable's AI platform, trained on an extensive database of claims and damaged property.
- North America > United States > California > San Francisco County > San Francisco (0.05)
- Asia > Middle East > Israel > Tel Aviv District > Tel Aviv (0.05)
- Asia > Japan (0.05)
How AI could be used in disaster preparedness, recovery
Imagine a hurricane has just devastated your town – and flattened your home. Before you could even begin to consider rebuilding, you'd wait weeks – or even months – for a property assessor just to visit to take a look at the damage, let alone unlock the funds you need to get back on your feet. But what if you could take pictures of the damage on the day of the hurricane, using your smartphone, and upload them – and when your insurer receives them, they deposit the funds into your bank account on the same day? If this sounds like the future, it's not – it's happening now. Real-world impact In September 2021, we at Tractable launched our estimating technology for the home, in Japan.
- Asia > Japan (0.28)
- Oceania > Tonga (0.05)
- North America > United States > California (0.05)
Root taps Tractable for AI claims platform
Root Insurance, a usage-based auto insurtech, will use artificial-intelligence technology from Tractable in its claims process. The first implementation will be of Tractable's AI Subro product for subrogation. "Implementing AI effectively will set the next generation of insurers apart from incumbents by creating greater efficiencies and a frictionless experience for consumers," says Mark LeMaster, chief claims and service officer for Root, in a statement. Root could not be reached for comment by publication. "The industry-leading accuracy and breadth of Tractable's solutions made for a clear choice as our AI partner and serves us well in our mission to deliver world-class technology when it matters most for our customers," LeMaster added.