Despite advancements in grounded content generation, production Large Language Models (LLMs) based applications still suffer from hallucinated answers. We present "Grounded in Context" - a member of Deepchecks' ORION (Output Reasoning-based InspectiON) family of lightweight evaluation models. It is our framework for hallucination detection, designed for production-scale long-context data and tailored to diverse use cases, including summarization, data extraction, and RAG. Inspired by RAG architecture, our method integrates retrieval and Natural Language Inference (NLI) models to predict factual consistency between premises and hypotheses using an encoder-based model with only a 512-token context window. Our framework identifies unsupported claims with an F1 score of 0.83 in RAGTruth's response-level classification task, matching methods that trained on the dataset, and outperforming all comparable frameworks using similar-sized models.

As the spring planting season arrives in College Station, Texas, certified master gardener Mark Smith is thrilled that peace is in the air. This time last year, a loud buzzing noise began disrupting Smith's morning routine of checking on the peppers, tomatoes, herbs, and shrubs growing in his backyard. Several times an hour, an Amazon Prime Air delivery drone would noisily emerge about 800 feet away, just past a line of trees behind Smith's home. His neighbors began calling the fleet flying chainsaws. Smith, a retired civil engineer, preferred a different comparison: "It was like your neighbor runs their leaf blower all day long," he says.

I was understandably apprehensive when I first heard about "Grounded," a game from Obsidian. "Grounded" is stressful in ways similar to most survival games. You scavenge for resources, build shelters, and manage your hunger, health, thirst and limited stamina. Your foes are all gargantuan insects, and spiders are a frequent antagonist -- among the toughest to overcome.

It is a phrase that more and more executives and developers alike are hearing in stand-up meetings and the board room. Companies from Google to IBM to Microsoft, are investing millions of dollars and man-hours into the development of deep learning platforms and products and spending largely on advertising and marketing to show their efforts. While such developments promise very exciting ventures in the months and years to come, organizations and individuals need to remember that deep learning is still a field that is expanding and going through its growing pains. True success with deep learning depends on the how, the what and the why of its application across business lines. Dr. Sid J. Reddy reflects these exact sentiments in his article, "Deep Learning is only as good as its data."

If you agree with the clear majority of respondents (nearly 85%) of a recent Boston Consulting Group and MIT Sloan Management Review survey, then you too believe that artificial intelligence can help push your business to gain or sustain a competitive advantage. Yet, at the same time, we hear cries from those in the AI industry who feel that the capabilities as they stand today -- and into the foreseeable future – are largely overblown. So that begs the question; who are we to trust? It certainly puts CIO's and IT architects in a precarious situation on how to handle AI-focused projects. Do you believe those that insist advanced AI is going to revolutionize the business world?

The following article was written by Vincent Conitzer and originally posted in Prospect Magazine. Progress in artificial intelligence has been rapid in recent years. Computer programs are dethroning humans in games ranging from Jeopardy to Go to poker. Self-driving cars are appearing on roads. AI is starting to outperform humans in image and speech recognition.