Mind Matters Natural and Artificial Intelligence News and Analysis

Gary Smith

mock-code-for-an-ai-large-language-model-llm-that-could-inte-1039114635-stockpack-adobestock
Mock code for an AI Large Language Model (LLM) that could intelligently answer questions. LLMs are one of the most popular implementations of recent Artificial Intelligence research

LLMs Still Cannot be Trusted for Financial Advice

The limitations of Large Language Models (chatbots) are illustrated by their struggles with financial advice
The real danger remains: we may think computers are smarter than us and consequently trust them to make decisions they should not be trusted to make. Read More ›
close-up-of-an-ai-artificial-intelligence-gpu-chip-on-mother-693521803-stockpack-adobe_stock
Close up of an AI Artificial Intelligence GPU chip on motherboard. Concept of LLM, Large Language Models.

Large Language Models: A Lack-of-Progress Report

They will not be as powerful as either hoped or feared
LLMs are enormously expensive yet it is increasingly clear that they will never be reliable enough to be trusted in complex situations where mistakes are costly Read More ›
digital-data-stream-analysis-cyber-environment-visual-displa-1270682305-stockpack-adobe_stock
Digital data stream analysis cyber environment visual display technology abstract perspective

Machine Learning Algos Often Fail: They Focus on Data, Ignore Theory

Without a theory, a pattern is just a pattern
The data-mining algorithms may recognize that a statistical pattern has vanished but won’t know why because they do not understand it as a logical relationship. Read More ›
2-seifenblasen-stockpack-adobe-stock-66135919-stockpack-adobe_stock
2 Seifenblasen

Yes, the AI Stock Bubble Is a Bubble

It's unfolding the way a financial bubble typically does
Unlike the Internet, LLMs are not useful enough to prompt customers to pay prices that reflect the hundreds of billions of dollars needed to develop them. Read More ›
person-uses-laptop-interacting-with-ai-virtual-assistant-ai-1133554305-stockpack-adobe_stock
Person uses laptop interacting with AI virtual assistant. AI head graphic overlay laptop keyboard. Concept of AI prompt engineering, LLM. Person types on keyboard to communicate with virtual

Why LLMs Are Not Boosting Productivity

If LLMs were as reliably useful as economist Tyler Cowen alleges, businesses would be using them to generate profits faster than LLMs generate text. They aren’t.
So far, AI is dragging down economic growth by diverting so much human talent and natural resources away from more productive uses. Read More ›
statistical-illusion-in-tv-game-isometric-illustration-of-un-805370148-stockpack-adobe_stock
Statistical illusion in TV game. Isometric illustration of undecided businessman on stairs with open doors. Modern landing page of making decision.

Intelligence Requires More Than Following Instructions

Post-training improves the accuracy and usefulness of LLMs but does not make them intelligent in any meaningful sense — as the Monty Hall problem shows
The danger is not that computers are smarter than us but that we think they are and thus trust them to make decisions they should not be trusted to make. Read More ›
group-of-jurors-sitting-together-in-jury-box-during-trial-st-740094298-stockpack-adobe_stock
Group of jurors sitting together in jury box during trial

The Large Language Model (LLM) “Superpower” Illusion Dies Hard

Historic confirmation bias around ESP and spirit cabinets makes for an interesting comparison with the current need to believe in the abilities of LLMs
Someone blinded by confirmation bias might believe that robots and Go victories are evidence that ChatGPT would be a reliable lawyer. Read More ›
robotics-or-ai-artificial-intelligence-connecting-interactio-626535089-stockpack-adobe_stock
Robotics or ai artificial intelligence connecting interaction with human.Chat bot software network.big data and transfer protocol system.Neuralink with smart brain.ai generative technology

Why LLMs (chatbots) Won’t Lead to Artificial General Intelligence

The biggest obstacle is seldom discussed: Most consequential real-world decisions involve uncertainty
LLMs can’t help when key decisions don’t feature objectively correct probabilities but rather subjective probabilities that need interpretation. Read More ›
game-of-tic-tac-toe-with-cinnamon-and-tangerine-on-a-black-b-692743521-stockpack-adobe_stock
Game of Tic-tac-toe with cinnamon and tangerine on a black background

Some Lessons From DeepSeek, Compared With Other Chatbots

I tested OpenAI o1, Copilot, and Gemini Flash, along with DeepSeek, on a question about Tic-Tac-Toe
Here’s the problem with the use case for LLMs: If you know the answer, you don’t need to ask an LLM; if you don’t know the answer, you can’t trust one. Read More ›
man-working-with-stock-market-data-stockpack-adobe-stock-882662733-stockpack-adobe_stock
Man working with stock market data.

Sloppy Science is a Statistical Sin

Evidence of sloppy science encourages readers to wonder if the entire research project is compromised
Journals should require authors to make their data available online before publication and encourage others to try to replicate the research they publish. Read More ›
large-language-models-concept-illustration-ai-stockpack-adob-891995515-stockpack-adobe_stock
large language models Concept illustration Ai

AGI Is Not Already Here. LLMs Are Still Not Even Intelligent

Recent tests continue to show huge failures in comprehending common sense issues
In one test, for example — not understanding the situation — Open AI launched an extremely verbose and confused description of the situation. Read More ›
generative-ai-a-visually-striking-graphic-that-is-split-into-794726345-stockpack-adobe_stock
Generative AI : A visually striking graphic that is split into two halves. On one side, depict a traditional library or database to symbolize Large Language Models

Large Language Models (LLMs) Flunk Word Game Connections

Despite hype, ChatGPT and its competitors, in all their iterations, are still just text-generators based on statistical patterns in the text databases they train on
Identification of statistical patterns in text that Large Language Models (LLMs) do not understand is not going to give us AGI, let alone superintelligence. Read More ›
llm-ai-large-language-model-concept-businessman-use-tablet-a-886081817-stockpack-adobe_stock
LLM, AI Large Language Model concept. Businessman use tablet and laptop with LLM icons. A language model distinguished by its general-purpose language generation capability. Chat AI.

The Promise of Artificial General Intelligence is Evaporating

Revenue from corporate adoption of AI continues to disappoint and, so far, pales in comparison to the revenue that sustained the dot-com bubble — until it didn’t
Recognition is growing that fundamental challenges make LLMs unreliable. Increasingly expensive scaling will likely hasten the popping of the AI bubble. Read More ›
blue-and-green-turf-meets-futuristic-stadium-lights-as-footb-932319723-stockpack-adobe_stock
Blue and green turf meets futuristic stadium lights as football players clash, pixelated projectiles in flight, technology merging with old-school gridiron glory.

Do Fantasy Sports Tell Us Something About Artificial Intelligence?

My biggest takeaway from my own involvement is how well fantasy football illuminates some weaknesses of artificial intelligence (AI)
AI models are really good at finding patterns in data but they are bad at deciding whether these spotted patterns can be used to make reliable predictions. Read More ›
vintage-baseball-on-a-weathered-american-flag-stockpack-adob-876174378-stockpack-adobe_stock
Vintage baseball on a weathered American flag

The World Series of Coin Flips

Here we go again with the annual coin-flipping ritual known as the World Series
With the Yankees and Dodgers so evenly matched, this World Series will be somewhat like a coin-flipping contest. That's the paradox of luck and skill. Read More ›
businessman-forecasting-a-crystal-ball-stockpack-adobe-stock-87433639-stockpack-adobe_stock
Businessman forecasting a crystal ball

P-Hacking: The Perils of Presidential Election Models

History professor Alan Lichtman’s model uses 13 true/false questions reflecting likely voter interests. But some of them seem rather subjective
Lichtman’s Thirteen Keys to the White House prediction model for the presidency shows some evidence of p-hacking, as does Helmut Norpoth’s Primary Model. Read More ›
high-tech-american-election-coverage-in-a-futuristic-studio-972672861-stockpack-adobe_stock
High-tech American election coverage in a futuristic studio

Presidential Pundits—a P-Hacking Parable

In politics, as elsewhere, too many studies flop when other researchers attempt to replicate them with fresh data
Some prediction models were developed by well-intentioned researchers before the perils of p-hacking were clearly understood, hence the failures. Read More ›
large-language-model-ai-machine-learning-concept-brain-business-stockpack-adobe-stock
large language model AI machine learning concept brain business

A Sloppy “AI Scientist” Could Make the Science Crisis Much Worse

A research team claims to have developed the AI Scientist that “generates novel research ideas, writes code, executes experiments ...” Really?
It continues to astonish us how easily people can be persuaded that LLMs (chatbots) can do our thinking for us Read More ›
four-leaf-clover-stands-out-against-green-leaves-stockpack-adobe-stock
Four-leaf clover stands out against green leaves

Bad Luck Seldom Persists — But it Never Guarantees Good Luck

Many people embrace the fallacious law of averages in their daily lives when "regression toward the mean" is a more realistic picture
For example, the baseball player with the highest batting average in any season generally does not do as well the season before or the season after. Read More ›
baseball-with-good-luck-sign-are-on-green-grass-stockpack-adobe-stock
Baseball with good luck sign are on green grass

What the Luck? How Luck Matters to Olympic and Major League Wins

One way to think about the relative importance of skill and luck is to consider the consistency of the outcomes
Statistical analysis shows that in major sports, the players or teams that won most of their games benefitted to a surprising degree from good luck. Read More ›