27 Articles about Data Science & AI

The way we measure progress in AI is terrible Many of the most popular benchmarks for AI models are outdated or poorly designed. – MIT Tech Review  

Multimodal AI: The Future of Enterprise Intelligence? – Information Week  

A discussion of data labeling for AI geospatial intelligence – Space News  

18 Recent Articles about AI’s future

AI Research Roundup including multi-modal LLM reasoning, 3D mesh generation with LLMs, image editing with AI, training code LLMs, and more - Pat McGuinness

AI Definitions: Test-time training (TTT)

Quantum mechanics offers a theoretical foundation and analytical method for understanding these vulnerabilities rooted in deep learning models – Nature  

10 Critical AI Concepts Explained in 5 Minutes – KD Nuggets

The NRO—how it got here and it’s ongoing mission – Center for Strategic & International Studies

AI Definitions: Structured Query Language, SQL

Generative AI Is Still Just a Prediction Machine – Harvard Business Review

12 Articles about  AI & the Military

Liquid foundation models promise competition for LLMs—here's how - Diginomica 

What AI Will Do To Data Science – Hakernoon      

AI Definitions: Liquid Foundation Models (LFM)

Getting The Biggest ROI On Your Digital Twin – Semi-engineering

Researchers are using machine learning to improve upon a predictive model for measuring electrons inside the Earth's outer radiation belt – Phys.org  

Scaling Laws for Pre-training Agents and World Models – Arxiv

Palantir moves into military and spy work by striking deals with AI developers – Gizmodo

Is ‘Big AI’ beating 'small AI'—and what does it mean for the military? – Defense One

Chinese researchers develop AI model for military use on back of Meta's Llama - Reuters

AI Definitions: Agentic AI Agents  

Agentic AI: How Large Language Models Are Shaping the Future of Autonomous Agents – Unite AI

How agentic AI handles the speed and volume of modern threats – HelpNetSecurity

AI Definitions: Predictive analytics  

An opinion piece about “What AI Will Do To Data Science – Hackernoon  

NGA Eyes Multimodal AI in Next Phase of Geospatial Analysis - GovCIO 

Could AI make data science obsolete? - ZDnet

Really Giving Thanks

Thanksgiving is more than just a long holiday weekend for watching eight consecutive hours of football or finding a few shopping bargains. It’s a time to reflect on the remarkable blessings showered upon us. Think back over the year; You’ll remember how often a crisis loomed. Perhaps you’re facing just such a situation now. And yet, how many of last year’s potential disasters are still with you? There is much for which to be thankful, and Thanksgiving is the perfect time to remind yourself and your family of what you are truly grateful. May you and your loved ones escape the daily grind and spend quality time together.  

Stephen Goforth

18 Recent Articles about AI’s future

Unrolling a Person

Development is an interesting word derived from a linguistic root meaning “rolled” or “folded.” An envelope is a folded sheet of paper, and to develop is to “unroll” something that has been heretofore so tightly rolled that we could not see what it really was. After the child has grown up, we can say that she was that way from the very start. But when she was a child, it was anyone’s guess how she would turn out.

The particular individual is an entity that is both utterly unique and profoundly like others. In this paradox of sameness and difference, we are like leaves on a tree or waves on the ocean.

The path of development is the fishtailing course we follow as we let go of what we have been and then discover a new thing to become—only to let go of that in time and become something new. This is the Way of Transition, the way or path of life itself, the alternating current of embodiment and disengagement, expansion and contractions.

William Bridges, The Way of Transitions

AI Definitions: Test-time training

Test-time training (TTT) - An alternative to transformers (which have high energy demands), TTTs theoretically do not grow when processing additional data, as transformers do. TTTs encode the data into representations called weights, so that additional data does not increase the size of the model. In effect, it is nestling a neural network inside another neural network. This type of machine learning model is in its early development stages and is only now being tested.

More AI definitions here

19 Articles about the Business of Running an AI Company

Pokémon Go Players Have Unwittingly Trained AI to Navigate the World - 404Media

Meta forms product group to build AI tools for businesses - Axios 

OpenAI Is Paying Dotdash Meredith At Least $16 Million to License Its Content – Ad Week 

AI Investments Are Booming, but Venture-Firm Profits Are at a Historic Low – Wall Street Journal

There’s No Longer Any Doubt That Hollywood Writing Is Powering AI – The Atlantic 

Researchers have invented a new system of logic that could boost critical thinking and AI – The Conversation 

AI Companies Are Trying to Get MIT Press Books – 404Media  

Liquid foundation models promise competition for LLMs - here's how - Diginomica 

Google preps ‘Jarvis’ AI agent that works in Chrome – 9to5Google 

AI firms need media more than they admit – Axios

Agentic AI: How Large Language Models Are Shaping the Future of Autonomous Agents – Unite AI 

Wall Street Giants to Make $50 Billion Bet on AI and Power Projects – Wall Street Journal

Meta strikes multi-year AI deal with Reuters – Axios

Apple releases new preview of its AI, including ChatGPT integration – CNBC

One of the Biggest AI Boomtowns Is Rising in a Tech-Industry Backwater - Wall Street Journal

OpenAI is looking beyond Microsoft for its cloud computing needs. – The Decoder

Microsoft Has an OpenAI Problem – New York Mag

AI firms need to address security, open-source concerns: G42 exec – Semafor  

HarperCollins Confirms It Has a Deal to Sell Authors' Work to AI Company - 404Media

Blame is contagious

Blame is contagious, according to UCLA researchers. Even when we observe a public display of blame, we are likelier to do the same.

Volunteers were asked to read about a governor blaming others for a problem, while a different group read how the governor accepted personal responsibility for the crisis. Both groups then wrote about a failure in their own lives. Those who saw blame modeled for them were almost a third more likely to join the blame game and put the fault for their failure on someone else. However, the number of blamers dropped when volunteers first wrote down their core values.

The researchers theorized that a reminder of how to make wise choices made it less likely for individuals to feel the need to defend themselves by blaming others and more willing to take responsibility.  

A USC professor conducted similar experiences and concluded that publicly blaming of others dramatically increases the likelihood that the practice will become viral.

When leaders, parents, or even friends make a practice of blaming others for their failures, they are encouraging people in their circle of influence to do the same. People then become less willing to take risks, less innovative and less creative—and less likely to learn from their mistakes.

Blame creates a culture of fear.

Stephen Goforth

 

A Guide to Success

Psychologist Gerd Gigerenzer argues that much of our behaviour is based on deceptively sophisticated rules-of-thumb, or “heuristics”. A robot programmed to chase and catch a ball would need to compute a series of complex differential equations to track the ball’s trajectory. But baseball players do so by instinctively following simple rules: run in the right general direction, and adjust your speed to keep a constant angle between eye and ball.

To make good decisions in a complex world, Gigerenzer says, you have to be skilled at ignoring information. He found that a portfolio of stocks picked by people he interviewed in the street did better than those chosen by experts. The pedestrians were using the “recognition heuristic”: they picked companies they’d heard of, which was a better guide to future success than any analysis of price-earning ratios.

Ian Leslie writing in The Economist

25 Podcasts about Journalism

The 404 Media Podcast —  A journalist-owned digital media company exploring the way technology is shaping–and is shaped by–our world. 

The Digiday Podcast — A weekly show about subscriptions, commerce, the modern newsroom, content creation, audio, streaming, and more.  

Freelancing for Journalists - How to approach freelancing, covering topics ranging from how to get started and what to include in pitches, to how to negotiate rates. Each episode includes guests on different career paths, and who have a variety of perspectives.    

IRE Radio Podcast (Investigative Reporters and Editors) — Behind the story with award-winning reporters, editors and producers to hear how they broke some big stories.

It's All Journalism — The series talks to working journalists about how they do their jobs, the latest trends in journalism, and the changing state of digital media.  

Journalism History — A scholarly journal covering the history of mass media.   

The Journalism Salute — A spotlight on interesting and important journalists and journalism organizations.

The Kicker — This Columbia Journalism Review podcast explores serious and challenging topics related to journalism and media.  

Longform Podcast (longform.org) — A weekly conversation with a non-fiction writer on how they tell stories.

Media Voices — Major media industry news each week from three experienced freelance journalists. The focus is on the business side of media and its impact on journalists’ work.

On the Media — Produced by WNYC radio, this is a weekly investigation into how the media shapes our worldview. 

Reveal (The Center for Investigative Reporting) — A look at CIR’s investigative reporting, focusing on real-world impact—from civil and criminal investigations to new laws and policies, better-informed conversations and community-driven solutions.

Reuters Institute for the Study of Journalism — A discussion of the Institute's research on trends in media. Based at the University of Oxford, this think tank offers research on the future of journalism.  

The Tip Off — A behind the scenes look at standout investigative reporting from the journalists themselves. 

Two Writers Slinging Yang — Sports journalist Jeff Pearlman hosts this podcast that’s all about the writing. Guests come from a variety of places.  

WriteLane (Tampa Bay Times and Poynter) — Some episodes explore a piece of the writing process: finding ideas, interviewing, seeking structure. Others dive deep into a single story, breaking down the how and why. Some include interviews with other journalists. (not updated) 

Podcasts about journalists doing journalism:

I'm Not A Monster (BBC Panorama and FRONTLINE PBS) — “How did an American family end up in the heart of the ISIS caliphate? Over four years, journalist Josh Baker unravels a dangerous story where nothing is as it seems.” 

The Other Latif (Radiolab)  — “How did this nerdy suburban Muslim kid come to be imprisoned at Guantanamo Bay?”

The Nobody Zone (RTÉ in Ireland and Third Ear in Denmark)— “In a forgotten London underworld, a homeless Irishman kills multiple times without detection, unseen in a world where nobody seems to care.”

The Canary | Washington Post Investigates —  “Two women and a shared refusal to stay silent. A seven-part podcast hosted by investigative reporter Amy Brittain.”

My Mother’s Murder (Tortoise)  — “‘My Mother’s Murder’ is an investigation by Paul Caruana Galizia into the life and killing of his mother Daphne Caruana Galizia. It’s an examination of the arrogance of power and the vast big-money corruption in a modern European country.”

White Lies (NPR) — “In 1965, Rev. James Reeb was murdered in Selma, Alabama. Three men were tried and acquitted, but no one was ever held to account. Fifty years later, two journalists from Alabama return to the city where it happened, expose the lies that kept the murder from being solved and uncover a story about guilt and memory that says as much about America today as it does about the past.”

The Line (Apple) — “Explore the impact of the forever wars on the U.S. Navy SEALs through the lens of the Eddie Gallagher case.”

The Lazarus Heist (BBC)  — “‘Almost a perfect crime.’ The hacking ring and an attempt to steal a billion dollars. Investigators blame North Korea. Pyongyang denies involvement. The story begins in Hollywood.”

In the Dark (American Public Media) — “We investigate the case of Curtis Flowers, a Black man from Winona, Mississippi, who was tried six times for the same crime. Flowers spent more than 20 years fighting for his life while a white prosecutor spent that same time trying just as hard to execute him.”

AI Definitions: SQL

SQL - Structured Query Language (SQL pronounced ess-kew-ell or sequel) is the most widely used method of accessing databases. This programming language can be used to create tables, change data, find particular data, and create relationships among different tables. For data scientists, it is second in importance after Python. Similar in structure and function to Excel, SQL can work with Excel and is able to handle billions of rows in multiple tables and thousands of users can access this data securely at the same time.

More AI definitions here

Diagnosis: AI v Doctors

“A small study found ChatGPT outdid human physicians when assessing medical case histories.” In fact, “Doctors often were not persuaded by the chatbot when it pointed out something that was at odds with their diagnoses. Instead, they tended to be wedded to their own idea of the correct diagnosis. They didn’t listen to AI when told things they didn’t agree with … But there was another issue: Only a fraction of the doctors realized they could literally paste the entire case into the chatbot and just ask it to give a comprehensive answer." -New York Times