#ai

An LLM TDD loop

Here’s a Bash script that uses a TDD loop with an LLM to iteratively write Python code to pass tests.

· Link post  #ai
Progress on mind reading

Researchers have developed a portable, wireless EEG acquisition system called BrainGPT capable of converting thoughts into text without the need for an fMRI machine.

· Link post  #breakthrough #ai
How AI has impacted freelancing

Jobs most negatively impacted by AI since ChatGPT’s release: writing jobs (-33%), translation jobs (-19%), and customer service jobs (-16%).

· Link post  #ai
Plentiful, high-paying jobs in the age of AI

Economic history shows an expansion in the variety of tasks performed by humans, with new jobs continually emerging despite automation.

· Link post  #ai
Automated unit tests using LLMs at Meta

Meta has developed TestGen-LLM, a tool using LLMs to enhance human-written tests, ensuring test suite improvements by passing certain filters.

· Link post  #ai
Klarna's massive use of AI for customer service

Klarna has launched an AI assistant powered by OpenAI, which is active across 23 markets and offers support in over 35 languages.

· Link post  #ai
How Australian small businesses are using ChatGPT

Business owners are increasingly finding AI like ChatGPT valuable for tasks beyond basic applications, such as data visualisation and financial reporting, saving time on manual data processing.

· Link post  #ai
Better Call GPT, Comparing Large Language Models Against Lawyers

The paper compares Large Language Models (LLMs) with Junior Lawyers and Legal Process Outsourcers (LPOs) in terms of contract review performance.

· Link post  #ai
Robots that cook

Researchers have developed a system called Mobile ALOHA for mobile manipulation tasks that require both hands and the whole body, aiming to improve robotic mobility and dexterity beyond simple table-top manipulation.

· Link post  #breakthrough #ai
Predicting sex from retina photos

Clinicians developed a deep learning model that predicts reported sex from retinal fundus photographs without needing to code.

· Link post  #ai
Local inference with MLX and Apple silicon

Apple has launched a new machine learning framework specifically for Apple Silicon.

· Link post  #ai
Inflection points for AI to replace specific occupations

These researchers suggest each occupation has an inflection point after which AI improvements harm human workers’ prospects.

· Link post  #ai
GPTs trained on ancient knowledge as an interactive history teacher

Here’s a GPT trained on knowledge from 17th-century texts. So, it answers in historical style, including outdated scientific concepts.

· Link post  #ai
The “it” in AI models is the dataset

The author has been at OpenAI for a year and observed that generative models closely approximate their training datasets.

· Link post  #ai
Thoughts on AI regulation

Recently, the Biden administration announced its framework to manage the deployment of AI by executive order. The order throws a bone to both sides of the AI argument: in some regards, the administration is embracing AI; in others, it’s hampering it. Overall, I’m disappointed.

· Note  #ai
GPT-4 is good at navigating iOS and Android

This team taught GPT to navigate iOS and Android by sending it screenshots and giving it instructions.

· Link post  #ai
Fumbling AI regulation with executive order

President Biden released an executive order on AI development, drawing parallels to the early fears and regulatory considerations during the dawn of the microprocessor and internet, highlighting that past technological advancements were less hindered by government intervention.

· Link post  #ai
Preventing AGI with hardware limits

I don’t want to single out jrincayc because these arguments are common from those concerned about AGI, and the author does acknowledge many arguments against their recommended approach. Sharing nonetheless, because this post clearly demonstrates several of the problems with the anti-AI movement within tech.

· Link post  #ai
Mind reading music listeners

Researchers from UC Berkeley, Shanghai, and Osaka University have trained a computer to analyse brain activity while listening to music and recreate the song, including a recognisable version of Another Brick in the Wall.

· Link post  #breakthrough #ai #music
How AI might impact growth

Some analysts predict AI could enable a 30% annual growth rate in the US economy, but Tyler Cowen argues for more moderate estimates, expecting a boost of ¼ to ½ of a percentage point.

· Link post  #ai
Is AI coming for middle management?

Large organisations often function with a single directive mind, usually a CEO or equivalent figure, supported by middle managers.

· Link post  #ai
Stealing data from keystrokes recorded over Zoom with 93% accuracy

Researchers have developed a deep learning model, with 95% accuracy, that can extract data from keyboard keystrokes recorded by a microphone and 93% when recorded on Zoom.

· Link post  #security #ai
Using AI to predict infectivity and variant evolution of viruses

The study presents an AI model that can predict how viral variants affect protein–protein binding.

· Link post  #ai #breakthrough
Large language models, explained with a minimum of math and jargon

Large Language Models (LLMs) like ChatGPT work by predicting the next word in a sequence, using word vectors or lists of numbers to represent words; similar words have similar vectors, making them closer in an imagined “word space.”

· Link post  #ai
The productivity effects of ChatGPT for writing

Noy and Zhang conducted an experiment with college-educated professionals using ChatGPT for writing tasks.

· Link post  #ai