Artificial Intelligence (AI) has been making subtle yet significant inroads into the daily workflows of tech professionals. Despite the lack of mainstream media coverage, these transformative tools are reshaping how work is done, often with profound benefits to individual workers rather than firms. Here, we explore two illustrative accounts from Nicholas Carlini and Erik Schluntz,...
Category: AI
What is quantization of LLMs?
Quantization, a compression technique, has long been utilized in various fields to map high precision values to lower precision ones, thus making data more manageable and less memory-intensive[1][2]. The advent of Large Language Models (LLMs) has necessitated the adoption of such techniques due to the exponential increase in model parameters and the associated computational demands....
What is AI model collapse?
AI model collapse is a phenomenon in artificial intelligence (AI) where trained models, especially those relying on synthetic data or AI-generated data, degrade over time. This degradation is characterized by increasingly limited output diversity, a tendency to stick to “safe” responses, and a reduced ability to generate creative or original content[1]. The phenomenon has significant...
‘Intersectional hallucinations’: why AI struggles to understand that a six-year-old can’t be a doctor or claim a pension
When you go to the hospital and get a blood test, the results are put in a dataset and compared with other patients’ results and population data. This lets doctors compare you (your blood, age, sex, health history, scans, etc) to other patients’ results and histories, allowing them to predict, manage and develop new treatments....
The EU AI Act: Your Essential Guide to New Regulations
The EU AI Act is finally here! Published in the Official Journal of the EU on July 12, this groundbreaking regulation comes into force today, August 1, 2024. But don’t worry, the changes won’t hit you all at once. There’s a detailed schedule outlining who needs to comply and when. Key Points You Need to...
The Trend Towards Smaller Language Models in AI
The landscape of artificial intelligence (AI) is undergoing a notable transformation, shifting from the pursuit of ever-larger language models (LLMs) to the development of smaller, more efficient models. This shift, driven by technological advancements and practical considerations, is redefining how AI systems are built, deployed, and utilized across various sectors. The Shift in AI Model...
Understanding Tokens in AI
In the realm of Artificial Intelligence (AI), particularly in the context of Large Language Models (LLMs), tokens are the fundamental units of language that enable machines to comprehend and generate human-like text. This article delves into the world of tokens, exploring their definition, history, types, and significance in the AI landscape. What are Tokens? In...
FraudGPT and other malicious AIs are the new frontier of online threats. What can we do?
The internet, a vast and indispensable resource for modern society, has a darker side where malicious activities thrive. From identity theft to sophisticated malware attacks, cyber criminals keep coming up with new scam methods. Widely available generative artificial intelligence (AI) tools have now added a new layer of complexity to the cyber security landscape. Staying...
Embeddings: The Secret Sauce in AI’s Data Cuisine
In the world of artificial intelligence (AI), data is the main ingredient. But just like raw ingredients need preparation before they become a delicious meal, data must be transformed into a palatable format for AI systems, particularly large language models (LLMs), to "digest." This is where embeddings enter the kitchen, serving as the secret sauce...
Training AI requires more data than we have — generating synthetic data could help solve this challenge
The rapid rise of generative artificial intelligence like OpenAI’s GPT-4 has brought remarkable advancements, but it also presents significant risks. One of the most pressing issues is model collapse, a phenomenon where AI models trained on largely AI-generated content tend to degrade over time. This degradation occurs as AI models lose information about their true...
