OpenAI quietly releases GPT-4o update amid leadership turmoil
OpenAI quietly releases a cheaper, more powerful GPT-4o update as the company faces leadership exodus and fierce competition in the rapidly evolving AI landscape.Read More
Inside the Dark World of Doxing for Profit
From tricking companies into handing over victims’ personal data to offering violence as a service, the online doxing ecosystem is not just still a problem—it’s getting more extreme.
Humans change their own behavior when training AI
A new cross-disciplinary study by Washington University in St. Louis researchers has uncovered an unexpected psychological phenomenon at the intersection of human behavior and artificial intelligence: When told they were training AI to play a bargaining game, participants actively adjusted their own behavior to appear more fair and just, an impulse with potentially important implications …
Read more “Humans change their own behavior when training AI”
Tips for Tuning Hyperparameters in Machine Learning Models
If you’re familiar with machine learning, you know that the training process allows the model to learn the optimal values for the parameters—or model coefficients—that characterize it. But machine learning models also have a set of hyperparameters whose values you should specify when training the model. So how do you find the optimal values for …
Read more “Tips for Tuning Hyperparameters in Machine Learning Models”
Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling
Large language models are trained on massive scrapes of the web, which are often unstructured, noisy, and poorly phrased. Current scaling laws show that learning from such data requires an abundance of both compute and data, which grows with the size of the model being trained. This is infeasible both because of the large compute …
Read more “Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modeling”
Build custom generative AI applications powered by Amazon Bedrock
With last month’s blog, I started a series of posts that highlight the key factors that are driving customers to choose Amazon Bedrock. I explored how Bedrock enables customers to build a secure, compliant foundation for generative AI applications. Now I’d like to turn to a slightly more technical, but equally important differentiator for Bedrock—the …
Read more “Build custom generative AI applications powered by Amazon Bedrock”
Announcing LangChain on Vertex AI for AlloyDB and Cloud SQL for PostgreSQL
Among application developers, LangChain is one of the most popular open-source LLM orchestration frameworks. To help developers use LangChain to create context-aware gen AI applications with Google Cloud databases, in March we open-sourced LangChain integrations for all of our Google Cloud databases including Vector stores, Document loaders, and Chat message history. And now, we’re excited …
Read more “Announcing LangChain on Vertex AI for AlloyDB and Cloud SQL for PostgreSQL”
A US Judge Ruled That Google Is an Illegal Monopolist. Here’s What Might Come Next
Judge Amit Mehta’s ruling has triggered a potentially yearslong process to decide how to punish the company. For users, it could mean a future in which Google isn’t front and center everywhere.
AI for mental health screening may carry biases based on gender, race
A growing body of AI tools screen how people talk, searching for subtle changes that could indicate mental health concerns like depression or anxiety. A study finds that these tools don’t perform consistently across people from different genders and races.
Integrating Scikit-Learn and Statsmodels for Regression
Statistics and Machine Learning both aim to extract insights from data, though their approaches differ significantly. Traditional statistics primarily concerns itself with inference, using the entire dataset to test hypotheses and estimate probabilities about a larger population. In contrast, machine learning emphasizes prediction and decision-making, typically employing a train-test split methodology where models learn from …
Read more “Integrating Scikit-Learn and Statsmodels for Regression”