WashPost: The Right's Current Thing: Claiming Trump Probes Are A Distraction
The Washington Post's Philip Bump examines media coverage of Trump's various investigations. Read The Full Article.
The Positioning Of Generative AI As Human Replacement
My inbox these days is filled with a never-ending stream of press pitches positioning generative AI as a solution to…
The Mismatch Between Conflict Forecasts & The Needs Of Policymakers: Ukraine
I recently received a press blast from a high-profile conflict forecasting organization widely touted by the US Government that it…
Enhancing Cyber Threat Identification In Open-Source Intelligence Feeds Through An Improved Semi-Supervised Generative Adversarial Learning Approach With Contrastive Learning
Responding to the challenge of efficiently leveraging Open-source Threat Intelligence Feeds (OTIFs) to enhance organizational security, this paper presents an…
No Vector Databases & Embeddings Don't Mitigate LLM Hallucination
As the enterprise world has begun to aggressively adopt LLMs into real-world workflows, the once-obscure and quickly dismissed challenge of…
Portfolio Analysis In Python: The Impact Of The Ukraine War On European And American Companies
A fascinating look at modeling the impact of the invasion of Ukraine on financial portfolios. Read The Full Article.
The Dangers As Companies Outsource Customer Relations To LLM's: A Case Study Of A Well-Known Pizza Chain
Today's Wall Street Journal explores how a growing number of fast food chains are using AI-powered speech recognition to take…
Fox News Dominates "Gold IRA" Mentions On TV News
The Washington Post today examines the "gold IRA" industry in the conservative news ecosystem. Fox News is replete with such…
How Evolving Guardrails & RLHF Is Creating False Confidence In LLM Safety & Bias Issues
One of the more interesting aspects of safety and bias issues in current generation LLMs is the rate at which…
How "Sorry For The Inconvenience" Rose In Our Lexicon
When SWB collapsed earlier this year, its website displayed a message that has become a symbol of modern life: "Our…
How TV News Covered The Odisha Train Crash Last Month
The train accident in India's Odisha state last month killed more than 260 people and injured more than 1,000. Yet,…
How Two LLMs See The Biden, Trump, Putin and Zelenskyy's Administrations: The Biases Of Web-Scale Training Data
As LLMs are increasingly positioned as an alternative to traditional search engines, what are the perspectives on various presidential administrations…
The Unintended Consequences & Harms Of Multimodal LLM Debiasing: Detection Vs Generation
Multimodal LLMs represent uncharted new territory in the push to "debias" and "globalize" computer vision models. Past generations of object…
Automated Image Captioning: Experiments With Google's New Imagen on Vertex AI Generative AI Service
With the release into general availability of Imagen on Vertex AI, Google's new image-based generative AI service, let's explore how…
When Ukraine Praised Russia For "Delivering" Weapons: LLMs & The Severe Risks Of Geopolitical Hallucination & Conflation
One of the more informative findings from yesterday's ambiguous image captioning experiment was the degree to which, when asked to…
LLMs & An Ambiguous Photo Caption From The Ukrainian Front Lines: Do LLMs Actually Reason Or Merely Pattern Match Their Training Data?
As we continue to explore the ability of LLMs to summarize and distill the chaotic conflicting cacophony that is the…
WashPost: James Comer Has A James Comer Problem
The Washington Post's Philip Bump includes an examination of media coverage of James Comer. Read The Full Article.
Large-Token LLMs: Leveraging Anthropic's Claude 2's 100K Token Limit To Summarize An Entire Episode Of Russia's 60 Minutes
Earlier today we demonstrated the extraordinary power of Anthropic's Claude 2's 100,000 token limit to summarize and topically annotate into…
Large-Token LLMs: Leveraging Anthropic's Claude 2's 100K Token Limit To Summarize An Entire ABC Evening News TV News Broadcast In A Single Prompt
One of the greatest limitations of current LLMs, besides hallucination and instability, is their severe input limits: most common production…
Experiments With Anthropic's Claude 2 For Summarization, Event & Relation Extraction, NER & Q&A
Continuing our explorations of how various commercial LLMs perform on the chaotic conflicting cacophony that is global news, what does…
Experiments With Google's PaLM 2 LLM's For Embedding, Summarization, Event & Relation Extraction, NER & Q&A: Bison & Gecko
What does it look like to summarize a television evening news broadcast using Google's PaLM 2 large language model (LLM)?…
Multimodal LLMs: What Bard Teaches Us About Consumer Versus Enterprise Applications & Imagery
Last week we explored some of the significant limitations of state of the art multimodal research LLM's when applied to…
Using Embeddings To Rank Clinical, "Creative" & "Inspired Fiction" Summaries Of An Evening News Broadcast
Earlier this month we demonstrated the use of embeddings to combat hallucination in LLM summarization and as a form of…
AGI & LLM Reasoning: Why Benchmarks Should Require 3-5 Runs To Reduce Anthropomorphization & False AGI Claims
Earlier today we examined how a SOTA multimodal LLM describes a variety of news and other images, testing how well…