Translated's Research Center

AI + Innovation

Updated weekly (last update 06/11/24)


When robots can’t riddle: What puzzles reveal about the depths of our own minds
AI might outsmart us in complex computations, but it still stumbles on simple riddles. Exploring how machines and humans solve puzzles not only pushes AI’s limits but also reveals the mysteries of our own minds. Could understanding AI’s shortcomings unlock the secrets of human reasoning?
Read the full article on BBC

Do AI models produce more original ideas than researchers?
Can AI generate more original ideas than researchers? A recent study by arXiv compared the work of 100 independent scientists with the 4,000 papers produced by Claude 3.5, which were then evaluated by neutral reviewers. The outcome might surprise you!
Read the full article on Nature

A day in the life of the world’s fastest supercomputer
It’s widely acknowledged that the majority of resources used in developing AI models have been allocated to the industrial sector. In this scenario, what role does the academic world play? The lack of technological resources has hindered the development of open-source models, but the new project, made possible by the world’s fastest computer, could potentially revolutionize the situation. Frontier is the key to democratizing AI: discover how it works. 
Read the full article on Nature

A new ‘AI scientist’ can write science papers without any human input. Here’s why that’s a problem
Is AI able to produce scientific papers? According to Sakana AI Labs, it can. In fact, recently was announced the creation of an AI scientist, an artificial intelligence system they claim can make scientific discoveries in the area of machine learning in a fully automated way. What perspectives does it open in research and is it a sustainable model for scientific development?
Read the full article on The Conversation

OpenAI is taking on Google with a new artificial intelligence search engine
OpenAI is releasing SearchGPT, a new tool able to facilitate the searching process of the users. In fact, differently from ChatGPT, the new engine permites to provide the most recent data and the sources. The OpenAI search engine could cement generative AI  as the future of finding answers online, after Google and others have experimented with early efforts to incorporate chatbots and AI-generated answers into the search experience.
Read the full article on CNN

‘A pivotal moment in neuroscience’: Scientists finally discover the brain cells that make you unique
Scientists have long been intrigued by the immense complexity of the human brain. Now, an international team of researchers has mapped its cellular composition, uncovering over 3,000 distinct cell types, including hundreds previously unknown.
Read the full article on BBC Science Focus

Putting Data Centers in Space Could Reduce Their Carbon Footprint, European Study Finds
Due to the environmental impact of AI, the European Union founded a study about the footprint of Earth-based data centers compared to Space-based ones. The study reveals that creating data centers in space can reduce the environmental impact of AI, and still can produce profit.
Read the full article on Wall Street Journal


AI-Driven Behavior Change Could Transform Health Care
Sam Altman – founder of ChatGPT – is promoting a new hyper personalized AI model which can help you to embrace a healthier lifestyle. In a world where more and more people suffer from chronic diseases, AI could democratize well-being with a personal tutor which helps you day by day to make the right decisions for your physical and emotional health. 
Read the full article on Time

Three ways AI is changing the 2024 Olympics for athletes and fans
In these Olympic games for the first time in history AI technologies are going to be widely used. This new technology will change many things both for athletes and viewers. From digital assistants to control athletes performances and routines, Paris Olympics are a good example of the importance of AI in diverse contexts.
Read the full article on Nature

We Asked AI to Take Us on a Tour of our cities. It Was Chaos. 
Littlefoot is an AI powered local discovery chatbot that can generate experiences in 161 cities around the world. Two Wired’s reporters tried to plan a perfect day in their own cities, London and New York, under 100$. The results weren’t as expected.
Read the full article on Wired

AI Needs Enormous Computing Power. Could Light-Based Chips Help?
The International Energy Agency forecasts that AI will consume ten times more power in 2026 than in 2023, with data centers using as much energy as the entire nation of Japan. To address this, optical computers, which use photons instead of electrons, could provide a more efficient option for AI tasks. This problem may have an unexpected solution: light-based chips. This article analyzes the pros and challenges of the new technology.
Read the full article on Quanta Magazine

The UAE Is on a Mission to Become an AI Power
An in-depth analysis of the UAE’s journey to becoming an AI power. Since it nominated the world’s first AI minister in 2017, the country has improved its technological assets and international partnerships (especially in the Global South). After the first Arabic open source model, the country wants to open up to the semiconductor market. It aims to find the best international scientists to work on UAE developments. It is interesting to note that the UAE has a very proactive and positive approach to these technologies, since they lead to economic diversification and social progress.
Read the full article on Time

An AI-based second opinion service could improve clinical decision-making today
A potential application of AI in medicine today: after an initial diagnosis, the doctor consults systems based on large language models (LLMs) to confirm the diagnosis or improve its precision.
Read the full article on STAT

Integrating AI into Products and Teams — with Disheng Qiu (Translated)
In this podcast episode, Disheng Qiu, VP of Engineering at Translated, delves into the challenges and costs of fine-tuning AI models, how to marry cutting-edge research and consumer products, the role of AI in enhancing productivity, and the Translated model, which focuses on humans-in-the-loop and continuous improvement. 
Listen to the Refactoring podcast

The digitalization of development
Major ministries and offices are increasingly relying on digital technologies and artificial intelligence. The perspective of balancing the dignity of human labor with digitalization is interesting; technology enhances human well-being (a very topical debate). “The process of digital transformation is no longer a matter of entertainment; it has become a lifeboat, and everyone must get on board, or they will find themselves outside the territorial boundaries of life’s ports.” In addition to the necessity of digitalization itself, there is a comment on the importance of familiarizing communities with these new technologies.
Read the full article on Al Sabah

What do a leafless tree, a fingerprint, and a harp have in common?
Alphabets represent a finite, standardized means of communication. The different emoji categories available on our keyboards have reached or are about to reach the same level of completeness as a regular alphabet. So what are the criteria to add more?
Find out in this article on Unicode

The sounds of invisible worlds
As with the invention of the microscope, the possibility of listening to sounds never heard before through the latest technologies has led to new discoveries.
Read the full article on Noēma Magazine

Robot trained to read braille at twice the speed of humans
A robot has been trained to read braille at the University of Cambridge, and it does so twice as fast as an average human and with a 90% accuracy rate. 
Read the full article on Science Daily

Running up that hill: Creativity, AI, and the human pursuit of uphill thinking
Vice President of Design and AI, John Maeda discusses key insights from his Design in Tech Report, reflects on the principles of designing with artificial intelligence, and advocates for embracing uphill thinking in a world that often prioritizes shortcuts.
Read the full article on Shortcut

AGI will not happen in your lifetime
An interesting discussion between cognitive scientist and psychologys Gary Marcus and software engineer Grady Booch discuss whether Artificial General Intelligence will be reached in the near future.
Read the full article on Substack

Artificial General Intelligence is already here
In this essay, authors Blaise Agüera y Arcas and Peter Norvig define AGI (Artificial General Intelligence) and argue that current applications of large language models already constitute examples of it.
Read the full article in Noēma Magazine

Visualizing AI
What does AI look like? DeepMind has asked 13 international artists to represent AI through visual arts pieces influenced by their conversations with scientists, engineers and ethicists.
Read more about it on DeepMind

Generative AI exists because of the Transformer
A useful article that provides a simple introduction to how LLMs work, a recommendable read for anyone that wants to learn more about this new technology.
Read the full article on the Financial Times

Meet Aleph Alpha, Europe’s answer to OpenAI
The European Union wants its own AI, and the German startup might be its best option. This could ease the regulatory process as the European bloc would be controlling “in-house” technologies – just one of several reasons why many in Europe are rooting for Aleph Alpha.
Read the full article on Wired

Revolutionary Mathematics: Artificial Intelligence, statistics, and the logic of capitalism
A podcast episode that focuses on the statistical models underpinning AI and financial systems and analyzes their potential to achieve significant goals. 
Listen to The Analysis News

AI is killing the old web, and the new web struggles to be born
The old web was already dying with the advent of apps and content that significantly lowered the attention span on selected web pages, but now AI is overwhelming the internet’s capacity to scale. James Vincent starts a crucial discussion.
Read the full article on The Verge

AI is a lot of work
Artificial Intelligence learns by finding patterns in data, but before that, this data must be put in order by humans, the workforce behind the machines. In this article, Josh Dzieza dives into the tedious work that lies behind enabling machines by themselves.
Read the full article on Intelligencer

OpenAI’s new tool attempts to explain language models’ behaviors
How do the LLM behind AI behave and why? It’s a question that has often remained unanswered. OpenAI is in the process of developing a tool that explains why a model reacts in a particular way.
Read the full article on TechCrunch


Inside the secret list of websites that make AI like ChatGPT sound smart
In the last few months, chatbots have exploded in popularity due to their high level of sophistication and precision when it comes to information. But where does this information come from?
Read the full article on The Washington Post


Big Tech teches AI
In this paper, Inga Luchs and her colleagues reveal how online machine learning courses offered by Google and IBM help consolidare the companies’ power. These big players stay ahead of the game by recruiting the top talents they teach, securing their future in the industry.
Read the paper on Sage Journals