Good morning. There is, as usual, a lot going on today.

But before we get into it, hear the words of famed thespian Nick Cage: “I am a big believer in not letting robots dream for us … The job of all art … is to hold a mirror to the external and internal stories of the human condition through the very human thoughtful and emotional process of recreation. A robot can’t do that. If we let robots do that … there will be no human response to life as we know it. It will be life as robots tell us to know it.”

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🛰️ AI for Good: An astronaut’s assistant

• 💻 OpenAI launches new research ‘agent’

• 🏛️ What Trump tariffs might mean for the tech sector

• 🌎 EU wades into open source as AI Act enters into force

AI for Good: An astronaut’s assistant

Astronauts aboard the International Space Station (ISS) face a number of challenges. They have a specific (and limited) timetable to conduct hundreds of experiments, all in rather close quarters. And they have to do it all in zero gravity. 

So Airbus, IBM and the German Aerospace Center teamed up to develop something they called Project Cimon, which stands for the “Crew Interactive Mobile Companion.” 

• This free-floating, generative AI-powered robot — built using 3D printing technology — was initially launched in 2018. 

• The idea was simple: operating through voice commands, Cimon would be able to assist the astronauts with experimental details while their hands were otherwise preoccupied. It’s notable that this AI assistant came about years before the recent AI craze. 

Cimon has been undergoing tests and evaluations ever since, mostly recently in 2024, when NASA’s Expedition 71 crew worked to demonstrate the efficacy of the Cimon assistant in aiding scientific research while reducing the crew’s workload.  

This is the Biggest Breakthrough in Next-Gen Tech Since iPhone—but hurry, their round is closing in less than a week.

Elf Labs, a next-gen tech entertainment company—with rights to billion-dollar character IP, like Cinderella, Snow White, Sleeping Beauty, and The Little Mermaid—has opened up an investment opportunity to the public, but the round closes soon.

With 100+ trademarks, 400+ copyrights, and powered by 12 revolutionary tech patents, Elf Labs is leading the way in immersive entertainment.

• Think hyper-realistic AI powered 3D worlds at scale. 

• Think virtual reality (VR)—without headsets. 

• Think global licensed merchandise ($350B industry).

Time is Running Out.  With just a week to invest in Elf Labs, the opportunity to become a part of this game-changing movement ends on February 12th.

OpenAI launches new research ‘agent’

The news: OpenAI on Monday launched a new agent, this one called “Deep Research,” which, in OpenAI’s words, works like this: “you give it a prompt, and ChatGPT will find, analyze and synthesize hundreds of online sources to create a comprehensive report at the level of a research analyst.” 

It is currently available only to Pro users ($200 per month), but will roll out to other tiers at a later date. It’s powered by a version of o3 that is optimized for web searching, and, according to OpenAI, is designed to include citations throughout its end result. 

The details: Unlike ChatGPT’s normal interface, the idea here isn’t real-time conversations. OpenAI said DeepResearch could take anywhere from five to 30 minutes to generate its report. 

• Unsurprisingly, OpenAI’s blog regarding the release was full of testimonials regarding DeepSeek’s capabilities, alongside high, but unverified, benchmark scores. 

• But the post also mentioned something that OpenAI and the like have been slipping into the fine print for two years now: “It can sometimes hallucinate facts in responses or make incorrect inferences … It may struggle with distinguishing authoritative information from rumors, and currently shows weakness in confidence calibration, often failing to convey uncertainty accurately.”

An example of the agent’s work on tariffs began circulating Sunday, eliciting this reaction from academics on Reddit who reviewed it: “I do worry it’s getting close enough where you’d get thousands of disinterested medical scientists around the world, who have to publish something, anything, for career reasons to use this to basically flood the zone with low-quality outputs.”

The landscape: I will note that Wharton professor Ethan Mollick thinks Deep Research is “very, very good.” Still, as Gary Marcus (unsurprisingly) wrote in response to Deep Research: because the results look good at a quick glance, “people will assume it’s legit, even when it isn’t.” 

• If actual researchers start using it, “we are all in trouble,” not just because of nuanced mistakes from hallucinations, but also from the potential for the skills necessary for research to atrophy.

• Mollick said in a separate post that professors should not allow their students to use Deep Research to prevent them from losing the skill involved in preparing research papers.

The other element to this involves energy intensity and associated carbon emissions, something OpenAI (also unsurprisingly) did not return my request for comment about. 

A ChatGPT query uses roughly 10x the amount of energy as a Google Search. But, since that number came out, ChatGPT queries have gotten more complex; as Dr. Sasha Luccioni recently pointed out regarding DeepSeek, despite training efficiencies, the model uses chain-of-thought ‘reasoning’, so its inference process (the usage of the model) is longer and less energy efficient. Here, with systems that are taking “tens of minutes” to generate a report, we can assume the resulting energy impact is significantly larger.

But, due to OpenAI’s lack of transparency, we do not know. 

Grab USA’S #1 beer machine for FREE!

With The Pinter—the all-in-one home brewing machine—you can enjoy 12 pints of delicious, award-winning beer with each brew, straight from your home.

Exclusive The Deep View special - get a FREE Pinter & save $174!

Buy your choice of 2 beer packs on our completely flex delivery plan and get:

• FREE Pinter machine ($149rrp)

• FREE tasting glass

• FREE delivery today!

Don’t miss out - join over 100,000 beer enthusiasts worldwide who’ve made the 

Pinter their go-to for fresh beer.

But hurry - this special deal is only on while stocks last! Order now to save up $174 and lock in your Pinter for FREE.

Claim offer now!

• Softbank, a newfound partner of OpenAI’s, said Monday that it would buy $3 billion worth of OpenAI’s products annually.

• Last week, a new platform — Compute Exchange — went live, promising an open and transparent means of buying and selling access to AI compute.

• Chinese AI app DeepSeek was downloaded by millions. Deleting it might come next (CNBC).

• That AI-restored Beatles song won Grammy for Best Rock Performance (The Verge).

• The LA fires have a shocking price tag — and we’ll all have to pick up the tab (Vox).

• Trump orders the creation of a U.S. sovereign wealth fund. Here's what that means (CBS News).

• Rubio says he’s acting director of USAID as humanitarian agency is taken over by the State Department (CNN).

What Trump tariffs might mean for tech and AI

The news: President Trump imposed tariffs of 25% against Canada and Mexico and 10% against China, a campaign promise that, on Monday, injected a healthy dose of chaos into the markets. The major indices initially tumbled but rebounded somewhat after the tariffs against both Mexico and Canada were paused for at least a month.

Trump said that the European Union will face tariffs next.

What it means: When and if the tariffs are either un-paused, or expanded to other countries, the resulting global trade war “will lead to weaker GDP growth, higher unemployment, higher interest rates and higher inflation,” according to research from Oxford Economics.

• Beyond leading to more expensive goods across a wide range of categories (specifically including technology products, many of which are produced overseas), tech stocks tend to struggle in highly inflationary environments.

• Central banks tend to raise interest rates in an effort to slow inflation. If interest rates rise, the cost of borrowing money gets more expensive, but, more importantly, the long-term estimates of a company’s earnings will drop. And tech companies tend to achieve massive valuations based on expectations of future earnings, rather than current value (this is something we saw in 2022, when the tech sector tumbled mightily). 

Still, the Oxford economists don’t expect the Federal Reserve to raise interest rates. They do, however, anticipate that the U.S. central bank will pause its current easing cycle, holding interest rates higher for longer. 

The Fed’s movements aside, the AI sector is already bracing for impact, beginning with the semiconductors — like Nvidia and TSMC — which rely on global supply chains to assemble their GPUs. Since the AI sector is largely circular at this point (the labs buy compute from the cloud companies, which build data centers using chips from the semiconductors), we might see GPUs become even more expensive as the supply chain takes a hit, inciting a potential slowdown in the AI trade, and a reduction in valuations. 

Nvidia closed the day down nearly 3%. TSMC dropped nearly 5%. Apple fell around 4%.

You can read more about tariffs here. 

EU wades into open source as AI Act enters into force

The news: In the midst of a regulatory environment that leading tech developers — such as Meta and OpenAI — are disinterested in partaking in, the European Union has announced plans to build its own family of multi-lingual large language models. The big difference? These will be fully open-sourced. 

The details: The EU said in a statement that a cohort of 20 research institutions and companies across the union will work together to develop this family of open LLMs. The effort — nicknamed OpenEuroLLM — will be coordinated by Charles University. 

• The intention here is to “democratize access to high-quality AI technologies and strengthen the ability of European companies to compete on a global market,” according to the statement. 

• Details and timelines remain unclear, but the project aims to ensure that the models, the data and the software will be “fully open and can be fine-tuned and instruction-tuned for specific industry and public sector needs.” 

The project, according to the Commission, has a total budget of roughly $38 million, a mere fraction of the billions that have been flowing into U.S. AI developers for the past few years. The grant reportedly does not cover the cost of compute. 

The models will be designed for compliance with the EU’s AI Act, the first rules of which became enforceable over the weekend. 

The details: The AI Act, still the global standard for AI regulation, takes a risk-based approach to regulating the tech, prohibiting those models that pose an “unacceptable risk” and imposing guardrails and transparency requirements on systems that pose high and limited risks. 

• The Act formally entered into force in August, with several layers of compliance windows set to close through August of 2026. On Sunday, the first window of compliance — which covers model prohibitions and AI literacy obligations — closed. 

• The prohibited practices include emotion recognition in workplaces and schools, social scoring and AI-based manipulation and deception. 

Companies can now face fines for non-compliance, which can amount to $35.8 million or 7% of a company’s annual revenue, whichever is higher. 

The Act has received its fair share of criticism, both from tech executives who aren’t looking forward to dealing with the restrictions, as well as activists and ethicists, who have pointed out the contradictions, loopholes, and, notably, government exemptions that were baked into the law. 

The prohibited applications of AI that entered into force this weekend do not, for instance, pertain to law enforcement. 

The OpenEuroLLM project has been scoffed at for its lack of scale. The budget is almost laughably small compared to the tabs being run up by U.S. developers. But, especially with compute bills uncovered (and unknown) this might not be the softball that it sounds like. 

At this point, there are plenty of other systems out there — some fully or partially open-sourced — that the project’s participants can build off of. And if the EU is able to build and deploy a baseline, open-source foundation model at low cost, they would achieve two things: one, prove that you don’t need the kind of extreme allocation of funds that U.S. firms are touting, and two, that the EU doesn’t need Big Tech firms at all. 

Meta and the rest have engaged in a pretty regular war against Europe and its regulation, slowing their rollouts or completely blocking access in Europe to their latest, AI-based products and services. If Europe is able to achieve that by itself, the Big Tech players will lose what bargaining power they have, and Europe will simultaneously and cleanly avoid the risk of becoming overly reliant on Big Tech. 

We’ll see how it plays out, but it’s another feather in the open-source cap, another battering ram against Big Tech’s technological walls.

🤔 Your thought process:

Selected Image 2 (Left):

• “The water in the other image was too still.”

Selected Image 1 (Right):

• “The angle of the reflection of the trees in image 2 looks wrong. The trees seem to be leaning to the right but the reflection doesn't.”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your view on o3 mini:

42% of you either don’t use OpenAI’s products, or don’t use generative AI at all.

20% said it’s okay, 14% said it’s amazing and 14% said it sucks. So. Mixed bag.

I don’t use AI:

• “ALL AI's have the same problem getting users, THEY DON’T EXPLAIN HOW TO USE IT!!! Good lord, we are not all computer experts, TEACH US HOW TO USE IT AND WE’LL FLOCK TO IT!!!”

I think this says more about the rote applicability and efficacy of the systems than it does about peoples’ computer expertise. But an interesting perspective, nonetheless.

If you want to get in front of an audience of 200,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

_{*Pinter Disclaimer: Voucher applied at checkout. Valid for new customers only. You must be 21+ to claim this promotion and to use a Pinter. 2 free Pinter Packs are only available when you sign up for a subscription - you can unsubscribe any time after your next order from your online account.}