Home Tech/AIThe Download: replicating the initial stages of pregnancy in a laboratory, along with an explanation of AI parameters

The Download: replicating the initial stages of pregnancy in a laboratory, along with an explanation of AI parameters

by admin
0 comments
The Download: replicating the initial stages of pregnancy in a laboratory, along with an explanation of AI parameters

This is the latest edition of The Download, our weekday newsletter delivering a daily insight into the happenings in the tech world.

Scientists are achieving pregnancy in organoids with human embryos

At a glance, it resembles the onset of a human pregnancy: A spherical embryo attaches to the uterine wall, gripping tightly as the initial strands of a future placenta manifest. This process is known as implantation — the official start of pregnancy.

Yet none of this is taking place within a living organism. These visuals were recorded in a lab in Beijing, within a microfluidic chip, with researchers observing the event unfold.

In three recent publications from Cell Press, researchers announce what they consider the most precise attempts thus far to replicate the early stages of pregnancy in a laboratory setting. They’ve sourced human embryos from IVF clinics and allowed these to join with “organoids” composed of endometrial cells, which create the uterine lining. Discover more about their findings and what may follow.

—Antonio Regalado

LLMs possess a HUGE number of parameters. But what constitutes a parameter?

The parameters of a large language model (LLM) are frequently described as the controls that dictate its operation. Imagine a gigantic pinball machine directing its balls bouncing from one side to the other through billions of precisely arranged paddles and bumpers. Adjust those configurations, and the balls will perform differently.

OpenAI’s GPT-3, launched in 2020, had 175 billion parameters. Google DeepMind’s newest LLM, Gemini 3, is speculated to have at least a trillion — some estimate closer to 7 trillion — though the company remains silent on the specifics. (In the current climate of intense competition, AI companies have stopped disclosing details about the construction of their models.)

Nevertheless, the fundamentals of parameters and their role in making LLMs perform the remarkable tasks they do remain consistent across various models. Ever pondered what makes an LLM truly work — the essence behind the lively pinball-machine analogy? Let’s explore.

—Will Douglas Heaven

The implications of new legal obstacles for the future of US offshore wind

This new year brings fresh legal skirmishes for offshore wind energy in the US.

On December 22, the Trump administration announced a halt to the leases of five wind farms currently under development off the US East Coast. Developers were instructed to cease operations immediately.

The stated rationale? Concerns over potential radar disturbances caused by the turbines. However, this is a recognized issue that developers have been addressing with the government for years.

Companies have rapidly initiated lawsuits, with court proceedings potentially commencing as soon as this week. Here’s what this recent turmoil may signify for the struggling offshore wind sector in the US.

—Casey Crownhart

This article comes from The Spark, our weekly newsletter illuminating the technologies that could address the climate crisis. Subscribe to have it delivered to your inbox every Wednesday.

The essential reads

I’ve scoured the web to gather for you today’s most entertaining/important/alarming/intriguing stories concerning technology.

1 Google and Character.AI have reached a settlement regarding a teenager’s passing
This is one of five lawsuits that the companies have resolved this week relating to the tragic deaths of young individuals. (NYT $)
+ AI companions represent the final phase of digital dependency, and lawmakers are targeting this issue. (MIT Technology Review)

2 The foremost output of the Trump administration is online trolling
Witness the Maduro memes. (The Atlantic $)

3 OpenAI has introduced a new ChatGPT Health feature
It focuses on analyzing medical results and addressing health inquiries. (Axios)
+ AI chatbots often fail to provide adequate healthcare advice for numerous questions about women’s health. (New Scientist $)
+ AI companies have stopped warning users that their chatbots do not qualify as medical professionals. (MIT Technology Review)

4 The acquisition of Manus by Meta is under investigation by China
Delaying the purchase provides it with additional leverage in negotiations with the US. (CNBC)
+ What occurred when we evaluated Manus. (MIT Technology Review)

5 China is establishing humanoid robot training centers
To tackle a significant lack of the data needed to enhance their performance. (Rest of World)
+ The competition in robotics is driving a demand for training data. (MIT Technology Review)

6 AI remains far from automating our jobs
The technology simply isn’t sufficiently advanced yet — at least for now. (WP $)

7 Weight regain appears to occur within two years of stopping the injections
This is the conclusion of an analysis encompassing more than 40 research studies. However, a closer look reveals that it’s not all doom and gloom. (New Scientist $)

8 This Silicon Valley group is relying on algorithms to find romance
Which seems somewhat like a misguided quest. (NYT $)

9 Hearing aids are set to significantly improve
Of course, thanks to advancements in AI. (IEEE Spectrum)

10 The first fully AI-generated film is projected to premiere within three years
According to Roku’s founder Anthony Wood. (Variety $)
+ What processes do AI models use to create videos? (MIT Technology Review)

Quote of the day

“I’ve viewed the video. Do not trust this propaganda machine. ”

—Tim Walz, governor of Minnesota, reacts on X to the Department of Homeland Security’s assertion that ICE’s shooting of a woman in Minneapolis was warranted.

Just one more thing

Exploring the peculiar limbo faced by millions of IVF embryos

Countless embryos created through IVF remain frozen in time, preserved in cryopreservation tanks across the globe. The number is continually increasing due to technological advancements, the growing acceptance of IVF, and enhancements in its success rates.

Fundamentally, an embryo is just a small mass of a hundred or so cells. However, unlike other bodily tissues, it carries the potential for life. Many contend that this gives embryos a specific moral significance, one that necessitates unique protections.

The challenge lies in the fact that there is no universal agreement on what that significance truly encompasses. Thus, as these embryos linger in stasis, patients, healthcare providers, embryologists, and lawmakers must confront the critical question of what should be done with them. What role do these embryos play in our lives? Who should bear responsibility for them? Read the complete story.

—Jessica Hamzelou

We can still enjoy nice things

A spot for comfort, enjoyment, and distraction to uplift your day. (Have any suggestions? Send me a message or connect with me.)

+ I love learning about musicians’ top songs 🎶
+ Here are some great tips for optimizing solo travel.
+ Take a look at a few of the promising new books set to be released this year.
+ I could endlessly enjoy this spherical variant of Snake (thanks, Rachel!)

You may also like

Leave a Comment