Wednesday, October 8, 2025

Polaris of Enlightenment

Artificial Intelligence and the Power of Language

The future of AI

How the mastery of language may be driving emergent abilities in Large Language Models, and what this means.

Published 7 May 2024
– By Thorsteinn Siglaugsson

4 minute read
This is an opinion piece. The author is responsible for the views expressed in the article.

A few days ago, Jamie Dimon, CEO of JPMorgan Chase, said that the advent of artificial intelligence could be likened to the discovery of electricity, so profound would be the societal changes it brings about. Artificial intelligence is certainly nothing new in banking; it has been used for decades. However, what is driving the discussion about the impact of artificial intelligence now is the emergence of Large Language models like ChatGPT. This is the major change, not only in the corporate world, but also in everyday life.

The Large Language models are unlike other AI tools in that they have mastered language; we can communicate with them in ordinary language. Thus, technical knowledge is no longer a prerequisite for using artificial intelligence in life and work; instead, expressive ability and understanding of language are key. But the development of these models and research into them also vividly remind us how language itself is the true prerequisite for human society.

Theory of Mind: Getting Into the Minds of Others

Large Language models function in a different way from normal software because they evolve and change without the developers and operators necessarily foreseeing those changes. The ability to put oneself in the mind of another person has generally been considered unique to humans. This ability, known in psychology as “theory of mind,” refers to an individual’s ability to formulate a “theory” about what another person’s mental world is like. This ability is fundamental to human society; without it, it’s hard to see how any society could thrive. Here’s a simple puzzle of this kind:

“There is a bag filled with popcorn. There is no chocolate in the bag. Yet the label on the bag says “chocolate” and not “popcorn.” Sam finds the bag. She had never seen the bag before. She cannot see what is inside the bag. She reads the label.”

The question is, what does she think is in the bag? Of course, the right answer is that Sam thinks there’s chocolate in the bag, because that’s what the label says. When Michal Kosinski, adjunct Professor at Stanford University, tested last year whether the first language models could handle this task, the result was negative. GPT-1 and 2 both answered incorrectly. But then he tried the next generation of the model, GPT-3. And in 40% of cases, it managed this type of task. GPT-3.5 managed it in 90% of cases and GPT-4 in 95% of cases.1

Emergent Capabilities of Large Language Models

This capability came as a surprise, as nothing had been done to build theory of mind capability into the models. They simply acquired it on their own as they grew larger and as the volume of data they were trained on increased. That this could happen is based on the models’ ability to use language, says Kosinski.

Another example I stumbled upon myself by chance recently was when GPT-4 asked me, after I had posed a puzzle to it, whether I had tried to solve the puzzle myself. The models certainly ask questions all the time, that’s nothing new, they aim to get more precise instructions. But this question is of a different nature. I answered yes and also mentioned that this was the first time I had received a question of this kind from the model. “Yes, you are observant,” GPT-4 replied, “with this I am trying to make the conversation more natural.”

Does this new development mean that the artificial intelligence truly puts itself in the mind of others? Does it mean it thinks, that it has feelings, opinions, an interest in the viewpoints and experiences of others? Of course, we can’t draw that conclusion. But what this means is that the behavior of the models is becoming increasingly similar to how we use language when we interact with each other. In this sense, we could actually talk about the mind of an AI model, just as we use theory of mind to infer about the minds of other humans.

The Power of Language

The language models draw our attention to the importance of language and how it underpins our societies and our existence. We now have a technology that is increasingly adept at using language, which has the advantage of possessing vastly more knowledge than any individual could possibly acquire in a lifetime and which can perform tasks much faster. We can use this technology to greatly enhance our own productivity, our reasoning, and our decisions if we use it correctly. This way, we can use it to gain more leisure time and improve our quality of life.

The comparison to the discovery of electricity is apt. Some might even want to go further and liken this revolution to the advent of language itself, which could be supported by pointing to the spontaneous capabilities of the models, such as theory of mind, which they achieve through nothing but the very ability to use language. What happens then if they evolve further than us, and could that possibly happen?

The fact that artificial intelligence has mastered language is a revolution that will lead to fundamental changes in society. The challenge we now face, each and every one of us, is to use it in a structured way, to our advantage, and avoid the pitfall of outsourcing our own thinking and decisions to it. The best way to do this is to enhance our own understanding of language, our expressive ability, and our critical thinking skills.

 

Thorsteinn Siglaugsson

 


  1. Kosinski, Michal: Theory of Mind May Have Spontaneously Emerged in Large Language Models, Stanford 2023. https://stanford.io/4aQosLV

Thorsteinn Siglaugsson is a Icelandic economist, consultant and writer. Chairman of the Icelandic Free Speech Society. Author: "From Symptoms to Causes" (Amazon). Regular contributor to The Daily Sceptic, Conservative Woman and Brownstone Institute. Siglaugsson also writes on Substack.

TNT is truly independent!

We don’t have a billionaire owner, and our unique reader-funded model keeps us free from political or corporate influence. This means we can fearlessly report the facts and shine a light on the misdeeds of those in power.

Consider a donation to keep our independent journalism running…

Anthropic challenges Google and OpenAI with new AI flagship model

The future of AI

Published 30 September 2025
– By Editorial Staff
AI companies' race continues at a rapid pace, now with a new model from Anthropic.
2 minute read

AI company Anthropic launches Claude Sonnet 4.5, described as the company’s most advanced AI system to date and market-leading for programming. According to the company, the model performs better than competitors from Google and OpenAI.

Anthropic has released its new flagship model Claude Sonnet 4.5, which the company claims is the best on the market for coding. According to reports, the model outperforms both Google’s Gemini 2.5 Pro and OpenAI’s GPT-5 on several coding benchmarks, writes TechCrunch.

One of the most remarkable features is the model’s ability to work independently for extended periods. During early testing with enterprise customers, Claude Sonnet 4.5 has been observed coding autonomously for up to 30 hours. During these work sessions, the AI model has not only built applications but also set up database services, purchased domain names, and conducted security audits.

Focus on safety and reliability

Anthropic emphasizes that Claude Sonnet 4.5 is also their safest model to date, with enhanced protection against manipulation and barriers against harmful content. The company states that the model can create “production-ready” applications rather than just prototypes, representing a step forward in reliability.

The model is available via the Claude API and in the Claude chatbot. Pricing for developers is set at 3 dollars per million input tokens and 15 dollars per million output tokens.

Fast pace in the AI race

The launch comes less than two months after the company’s previous flagship model, Claude Opus 4.1. This rapid development pace illustrates, according to TechCrunch, how difficult it is for AI companies to maintain an advantage in the intense competition.

Anthropic’s models have become popular among developers, and major tech companies like Apple and Meta are reported to use Claude internally.

AI-created viruses can kill bacteria

The future of AI

Published 28 September 2025
– By Editorial Staff
Bacteriophages attach to bacteria, inject their DNA and multiply until the bacteria burst. AI can now design new variants from scratch.
2 minute read

Researchers in California have used artificial intelligence to design viruses that can reproduce and kill bacteria.

The breakthrough opens up new medical treatments – but also risks becoming a dangerous weapon in the wrong hands.

Researchers at Stanford University and the Arc Institute have for the first time succeeded in creating complete genomes using artificial intelligence. Their AI-designed viruses can actually reproduce and kill bacteria.

— That was pretty striking, just actually seeing, like, this AI-generated sphere, says Brian Hie, who leads the laboratory at the Arc Institute where the work was carried out.

The team used an AI called Evo, trained on genomes from around 2 million bacteriophages (viruses that attack bacteria). They chose to work with phiX174, a simple virus with just 11 genes and 5,000 DNA letters.

16 of 302 worked

The researchers let the AI design 302 different genome variants, which were then chemically manufactured as DNA strands. When they mixed these with E. coli bacteria, they achieved a breakthrough: 16 of the designs worked and created viruses that could reproduce.

— They saw viruses with new genes, with truncated genes, and even different gene orders and arrangements, says Jef Boeke, biologist at NYU Langone Health who was given advance access to the study.

Since viruses are not considered living organisms, this is not yet truly AI-designed life – but it is an important first step toward that technology.

Major medical potential

The technology has great potential in medicine. “Most gene therapy uses viruses to shuttle genes into patients’ bodies, and AI might develop more effective ones”, explains Samuel King, the student who led the project.

Doctors have previously tried so-called phage therapy to combat serious bacterial infections, something that AI-designed viruses could improve.

“Grave concerns”

But the technology’s development also raises strong concerns. The researchers have deliberately avoided training their AI on viruses that infect humans, but others could misuse the method.

— One area where I urge extreme caution is any viral enhancement research, especially when it’s random so you don’t know what you are getting. If someone did this with smallpox or anthrax, I would have grave concerns, warns J. Craig Venter, a pioneer in synthetic biology.

Venter believes that the technology is fundamentally based on the same trial-and-error principle that he himself used two decades ago, just much faster.

Future challenges

Creating larger organisms is significantly more difficult. E. coli has a thousand times more DNA than phiX174. “The complexity would rocket from staggering to way way more than the number of subatomic particles in the universe”. explains Boeke.

Jason Kelly, CEO of biotech company Ginkgo Bioworks, believes that automated laboratories where AI continuously improves its genome designs will be needed for future breakthroughs.

— This would be a nation-scale scientific milestone, as cells are the building blocks of all life. The US should make sure we get to it first, says Kelly.

China plans fully AI-controlled economy by 2035

The modern China

Published 26 September 2025
– By Editorial Staff
By 2035, AI is planned to have "completely reworked Chinese society" and implemented a new phase of economic and social production.
2 minute read

The Chinese government has presented an ambitious ten-year plan where artificial intelligence will permeate all sectors of society by 2035 and become the “main engine for economic growth”.

China’s State Council has published a comprehensive plan aimed at making the country the world’s first fully AI-driven economy within eleven years. According to the government document presented at the end of August, artificial intelligence will have transformed Chinese society by 2035 and become the foundation for what is described as “a new phase of development in intelligent economy and intelligent society”.

The plan, which spans ten years, encompasses six central societal sectors that will be permeated by AI technology by 2027. These include science and technology, citizen welfare, industrial development, consumer goods, governance, and international relations.

The goal: 90 percent usage by 2030

According to the timeline, AI technology should reach a 90 percent usage rate by 2030 and practically become a new type of infrastructure. At this point, the technology is expected to have developed into a “significant growth engine for China’s economy”.

The strategy resembles the country’s previous “internet plus” initiative, which successfully integrated the internet as a central component in the Chinese economy.

By 2035, AI should according to the plan have “completely reworked Chinese society” and implemented a new phase of economic and social production. This is an ambitious goal with significant consequences, not only for the People’s Republic but for the entire world.

International cooperation in focus

The State Council emphasizes that AI should be treated as an “international public good that benefits humanity”. The plan highlights the importance of developing open source AI, supporting developing countries in building their own technology sectors, and the UN’s role as a leader in AI regulation.

Although China’s AI industry is growing rapidly, as exemplified by the open AI platform DeepSeek’s successes earlier this year, Chinese models still lag several months behind their American counterparts in terms of average performance. This is largely due to restrictions and barriers that Western countries have imposed.

However, the gap is steadily narrowing. At the end of 2023, American AI models performed better than Chinese ones in 13 percent of general reasoning tests. By the same time in 2024, this figure had dropped to 8.1 percent. In certain AI applications, China is already a world leader and has invested heavily in offering its services at low prices and in many cases completely free as open source.

The State Council’s ten-year plan aims to further reduce the lead by strengthening key areas such as fundamental model performance, security measures, data access, and energy management.

Whether Beijing can deliver on its massive goals with the help of sometimes unreliable technology remains to be seen. However, if other nationally coordinated plans are any indication, the country may face a comprehensive transformation.

New robot takes on household chores

The future of AI

Published 7 September 2025
– By Editorial Staff
1 minute read

The AI robot Helix can wash dishes, fold laundry and collaborate with other robots. It is the first robot of its kind that can control the entire upper part of the body.

The American robotics company Figure AI’s new humanoid robot has visual perception, language understanding and full control over fingers, wrists, torso and head. This enables the robot to pick up small objects and thereby help with household tasks.

Helix is powered by a so-called dual-system architecture, which can be explained as having a unique “two-brain” AI architecture where one part interprets language and vision while another part controls movements quickly and precisely.

Among other things, the company demonstrates that the robot can load dishes into the dishwasher, fold laundry and sort groceries. The robot can also sort and weigh packages at postal facilities.

It can also handle thousands of new objects in cluttered environments, without prior demonstrations or custom programming. This means it can perform tasks it is not programmed for and is designed to solve problems independently in an unpredictable environment.

It can follow voice commands in a similar way to talking with a human and act accordingly. What also makes the robot special is that it can collaborate with other robots. In tests, for example, two Helix robots have successfully been able to work together to unpack groceries.

Our independent journalism needs your support!
We appreciate all of your donations to keep us alive and running.

Our independent journalism needs your support!
Consider a donation.

You can donate any amount of your choosing, one-time payment or even monthly.
We appreciate all of your donations to keep us alive and running.

Dont miss another article!

Sign up for our newsletter today!

Take part of uncensored news – free from industry interests and political correctness from the Polaris of Enlightenment – every week.