Regulating Generative AI - LLMOps pipelines with TransparencyDebmalya Biswas
The growing adoption of Gen AI, esp. LLMs, has re-ignited the discussion around AI Regulations — to ensure that AI/ML systems are responsibly trained and deployed. Unfortunately, this effort is complicated by multiple governmental organizations and regulatory bodies releasing their own guidelines and policies with little to no agreement on the definition of terms.
Rather than trying to understand and regulate all types of AI, we recommend a different (and practical) approach in this talk based on AI Transparency —
to transparently outline the capabilities of the AI system based on its training methodology and set realistic expectations with respect to what it can (and cannot) do.
We outline LLMOps architecture patterns and show how the proposed approach can be integrated at different stages of the LLMOps pipeline capturing the model's capabilities. In addition, the AI system provider also specifies scenarios where (they believe that) the system can make mistakes, and recommends a ‘safe’ approach with guardrails for those scenarios.
Regulating Generative AI - LLMOps pipelines with TransparencyDebmalya Biswas
The growing adoption of Gen AI, esp. LLMs, has re-ignited the discussion around AI Regulations — to ensure that AI/ML systems are responsibly trained and deployed. Unfortunately, this effort is complicated by multiple governmental organizations and regulatory bodies releasing their own guidelines and policies with little to no agreement on the definition of terms.
Rather than trying to understand and regulate all types of AI, we recommend a different (and practical) approach in this talk based on AI Transparency —
to transparently outline the capabilities of the AI system based on its training methodology and set realistic expectations with respect to what it can (and cannot) do.
We outline LLMOps architecture patterns and show how the proposed approach can be integrated at different stages of the LLMOps pipeline capturing the model's capabilities. In addition, the AI system provider also specifies scenarios where (they believe that) the system can make mistakes, and recommends a ‘safe’ approach with guardrails for those scenarios.
Generative AI by Salesforce Admin Group DehradunkailashChandra95
The document introduces Einstein GPT, the first generative AI for CRM. It begins with an agenda that outlines discussing essential AI terminology, what generative AI is, how CRM can use generative AI, and Einstein GPT. It then defines key AI concepts like artificial intelligence, neural networks, deep learning, natural language processing, generators, transformers, large language models, and generative pre-trained transformers. It explains that generative AI can quickly generate new content based on inputs. For CRM, generative AI can personalize emails, product descriptions, marketing pages, and customer service replies to make CRM more powerful.
DeepMind achieved multiple breakthroughs in 2021 related to our prediction, including:
- Proposing a method using neural networks and human collaboration to generate conjectures in mathematics. This led to solving a long-standing conjecture and proving a new theorem.
- Approximating the density functional theory in materials science using a neural network trained on mathematical constraints.
- Repurposing AlphaZero to discover new deterministic matrix multiplication algorithms by framing it as a reinforcement learning problem.
- Developing a deep reinforcement learning system to stabilize plasma in nuclear fusion experiments, bringing controlled fusion closer to reality.
This document discusses applying agile methods in the automotive industry. It notes the differences between logistics in startups versus large automotive companies. It suggests that agile can be used throughout the automotive organization, from product owners to CxOs to suppliers. Challenges include integrating agile with existing processes and standards in automotive like AUTOSAR and ensuring safety. Emerging technologies like electric vehicles, advanced driver assistance systems, and automated driving require new approaches.
The document discusses variable neighborhood search (VNS), a metaheuristic technique for solving optimization problems. It provides background on VNS, explaining that it was proposed in 1997 and explores predefined neighborhoods to escape local optima. The main concepts of VNS are described as involving shaking to move to a new neighborhood, local search to improve solutions, and moving to another neighborhood if no improvement is found. An algorithm is presented showing the process of defining neighborhoods, generating initial solutions, applying local search, and moving to new neighborhoods iteratively until a termination criteria is met. A variety of applications where VNS has been used are listed, such as scheduling, vehicle routing, and network design problems.
This session was presented at the AWS Community Day in Munich (September 2023). It's for builders that heard the buzz about Generative AI but can’t quite grok it yet. Useful if you are eager to connect the dots on the Generative AI terminology and get a fast start for you to explore further and navigate the space. This session is largely product agnostic and meant to give you the fundamentals to get started.
What's the difference between RPA and Intelligent Automation? RPA is an approach making use of software bots to automate mundane, high-volume, rule-based and repeatable tasks. Intelligent Automation is an advanced form of RPA. Check out our Infographics to understand the major differences.
Generative AI by Salesforce Admin Group DehradunkailashChandra95
The document introduces Einstein GPT, the first generative AI for CRM. It begins with an agenda that outlines discussing essential AI terminology, what generative AI is, how CRM can use generative AI, and Einstein GPT. It then defines key AI concepts like artificial intelligence, neural networks, deep learning, natural language processing, generators, transformers, large language models, and generative pre-trained transformers. It explains that generative AI can quickly generate new content based on inputs. For CRM, generative AI can personalize emails, product descriptions, marketing pages, and customer service replies to make CRM more powerful.
DeepMind achieved multiple breakthroughs in 2021 related to our prediction, including:
- Proposing a method using neural networks and human collaboration to generate conjectures in mathematics. This led to solving a long-standing conjecture and proving a new theorem.
- Approximating the density functional theory in materials science using a neural network trained on mathematical constraints.
- Repurposing AlphaZero to discover new deterministic matrix multiplication algorithms by framing it as a reinforcement learning problem.
- Developing a deep reinforcement learning system to stabilize plasma in nuclear fusion experiments, bringing controlled fusion closer to reality.
This document discusses applying agile methods in the automotive industry. It notes the differences between logistics in startups versus large automotive companies. It suggests that agile can be used throughout the automotive organization, from product owners to CxOs to suppliers. Challenges include integrating agile with existing processes and standards in automotive like AUTOSAR and ensuring safety. Emerging technologies like electric vehicles, advanced driver assistance systems, and automated driving require new approaches.
The document discusses variable neighborhood search (VNS), a metaheuristic technique for solving optimization problems. It provides background on VNS, explaining that it was proposed in 1997 and explores predefined neighborhoods to escape local optima. The main concepts of VNS are described as involving shaking to move to a new neighborhood, local search to improve solutions, and moving to another neighborhood if no improvement is found. An algorithm is presented showing the process of defining neighborhoods, generating initial solutions, applying local search, and moving to new neighborhoods iteratively until a termination criteria is met. A variety of applications where VNS has been used are listed, such as scheduling, vehicle routing, and network design problems.
This session was presented at the AWS Community Day in Munich (September 2023). It's for builders that heard the buzz about Generative AI but can’t quite grok it yet. Useful if you are eager to connect the dots on the Generative AI terminology and get a fast start for you to explore further and navigate the space. This session is largely product agnostic and meant to give you the fundamentals to get started.
What's the difference between RPA and Intelligent Automation? RPA is an approach making use of software bots to automate mundane, high-volume, rule-based and repeatable tasks. Intelligent Automation is an advanced form of RPA. Check out our Infographics to understand the major differences.