Ringing the DeepSeek alarm bells: What we're missing

March 10, 2025

Good morning. Coming to you live this week from the HumanX conference in Las Vegas.

Got a lot of interesting stuff on the way.

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

In today’s newsletter:

🛰️ AI for Good: Life beyond Earth
💻 Google’s push for AI dominance through Search
🚨 Guest Post — Ringing the DeepSeek alarm bells: What we're missing

Source: Unsplash

It is perhaps unsurprising that, in the search for intelligent, extraterrestrial life, scientists have turned to machine learning.

Here’s what’s going on: Project Galaxia, launched by an anonymous scientist and now supported by a coalition of scientific institutions, aims to leverage both machine learning and quantum cognition to “analyze vast amounts of astronomical data, interpret complex radio signals and uncover potential signs of life beyond Earth.”

The idea is that, somewhere in that vast trove of deep space data, machine learning models tuned and trained to detect anomalies and analyze patterns might discover “potential biosignatures, technological signals and other signs” that point to life beyond Earth.
The project’s findings are currently undergoing a “rigorous” peer review process from a team of scientists around the world.

According to the project, initial findings have “baffled” the project’s creators. And this year, Project Galaxia aims to release its first comprehensive analysis of all this data, making 2025 a big year in the hunt for alien life.

Ride top e-Bike brands like Aventon, Specialized, and Cannondale — all warrantied, and up to 60% off retail.

Upway makes buying (or selling) your e-Bike simple, with fast delivery and expert support.

Plus, get $150 off your next e-Bike with code thedeepview.

Don’t miss out — upgrade your ride today.

Source: Google

If I had to guess, I’d say that no one has quite forgotten Google’s rollout of AI Overviews, its integration of generative AI into its wildly popular search engine, an integration that led to some viral pieces of super sound advice: that people should put glue on pizza to keep the cheese from sliding off; that pregnant women should smoke and that people should eat rocks.

Google CEO Sundar Pichai acknowledged at the time that hallucinations were an unsolved problem, and Google took steps to address Overviews.

But it has not ceased its relentless, if more gradual, rollout of AI Overviews, a rollout that, according to Google, is finally going well — the tech giant said last week that Overviews are now used by more than a billion people (although it’s tough to tell how meaningful that metric is, considering the fact that it’s nearly impossible to turn Overviews off).

The news: At the same time, Google said that it is amping up Overviews’ capability by integrating its latest generative AI model — Gemini 2.0 — into Overviews in a clear challenge to AI Search competitors like Perplexity and OpenAI.

In addition to rolling out AI Overviews to more users — users no longer need to sign in to access Overviews — Google also unveiled an experiment it calls AI Mode, a search option that nixes the classic blue links entirely, replacing them with an interface that looks and functions quite similarly to Perplexity or OpenAI’s search.
It’s not clear when this will become widely available, though it seems likely that this will further challenge publishers trying to drive traffic to their websites through Google, a search engine with a greater than 90% slice of the search market.

“As with any early-stage AI product, we won’t always get it right,” Google said. “For example, while we aim for AI responses in Search to present information objectively based on what’s available on the web, it’s possible that some responses may unintentionally appear to take on a persona or reflect a particular opinion.”

Google said that it will soon begin inviting Google One AI Premium subscribers to test out the feature.

"2025 is the year of inevitability, the year everyone is going to be using AI for search and discovery," according to Jim Yu, the CEO and co-founder of SEO firm BrightEdge. “Google has long dominated the search landscape, and they're still the undisputed king in terms of market share today. Doubling down on the presence of AI Overviews, proves they're not afraid to aggressively fight for their position in this new arms race. The competition is fierce, the stakes are high and the future of search hangs in the balance.”

GenAI retrieval should be precise, secure, & scalable — without complexity.

The Coveo Passage Retrieval API integrates seamlessly with any LLM, so your AI applications stay flexible and future-ready.

Custom GenAI Experiences - Improve the reliability and security of any generative AI application by providing your LLM with the most relevant passages from unified enterprise content sources.
Chatbots & Virtual Assistants - Improve the relevance of the answers from your chatbots & virtual assistants so your users experience personalized, context-rich conversations.
Internal Knowledge Management - Empower employees with personalized, accurate answers grounded in enterprise knowledge, reducing search time and boosting productivity.
Customer-Facing Support & Case Deflection - Provide precise self-service answers cited from support knowledge that minimize escalations, lower support costs, and boost customer satisfaction.

Explore the API now.

The DeepSeek saga: The U.S. is considering banning DeepSeek’s models from government devices, according to the Wall Street Journal. The move would mirror similar actions that have already been taken by countries around the world, including Canada, Australia and South Korea. Officials are additionally considering banning DeepSeek from U.S. app stores.
Musk’s Colossus: Elon Musk’s xAI has acquired a one-million-square-foot property in Memphis designed to bolster its Memphis data center, a colossal cluster of 200,000 (+) Nvidia GPUs that is straining the local power grid and worsening local air quality.

OpenAI, Oracle eye Nvidia chips worth billions for Stargate site (Bloomberg).
AI copyright shake-up could breach international law (The Times).
SpaceX rocket explosion leads to flight delays at Florida airports over fears of falling debris (NBC News).
Why most countries are struggling to shut down 2G (Rest of World).
Scale AI is being investigated by the US Department of Labor (TechCrunch).

The following is a guest column written by Dr. Joe Sutherland, the first director of the Emory Center for AI Learning.

Source: Created with AI by The Deep View

Tech stock selloffs have hit your retirement account. Trade tensions have increased the prices you pay for housing and food. Fears abound that your private data will be leaked to bad actors overseas who will use it to impersonate you. Unless you’ve been living under a rock, you’ve likely heard about DeepSeek, the Chinese startup entangled in a chain of events contributing to our increasingly volatile everyday lives.

In January, DeepSeek released an AI application called R1 that was reported to have achieved a level of performance equal to that of industry leaders like Open AI’s ChatGPT and xAI’s Grok. Quickly demo R1, which is available freely online, and you will find that its performance really does feel just as good as, if not better than, a ChatGPT, a Claude, or a Grok. That’s a problem so big that federal bills have been proposed to make using DeepSeek illegal.

A Wakeup Call for the West

The present debate in the popular media focuses on competition between the United States and China: perhaps the United States doesn’t have the technological advantage we thought it had.

The fact that a start-up out of China was able to create something that performs at state-of-the-art levels is a wake-up call for folks who thought Western bloc technologies would remain dominant for the foreseeable future. That raises the same questions that the United States started asking when the Soviet Union sent Sputnik on its maiden orbit: Do they know something that we don't? If they possess this advantage, what other advantages do they possess that we don’t know about? The obvious take-home here is that the United States isn't necessarily winning the analogous "space race.”
But that take-home obscures the flaws in our fundamental assumptions about what will make the United States competitive in artificial intelligence. The blockbuster revelation from DeepSeek’s release of R1 was this: it only took DeepSeek’s developers a couple million dollars to train an industry-leading model.

In other words, a little startup — which was late to the game — was able to release a multi-billion-dollar product for less money than these big tech firms spend on just one or two senior engineers.

Punching Holes in Big Tech's Approach

The big tech firms have been raising billions of dollars of capital (see, for example, Stargate) by arguing that the most effective way to assert dominance in AI is to build bigger, more complex models — complete with bigger datasets, more advanced chips, more electricity, greater real estate for data centers and top engineering talent.

The DeepSeek revelation throws a wrench in the story leading firms have been telling: that success in LLMs is capital intensive, that bigger is better and that early winners will dominate. And they’ve told that story at the expense of supply-constrained resources — like chips, talent and energy.

Speed v. Efficiency

If a model like DeepSeek can use clever mathematical optimizations — the bulk of which were readily accessible in the public scientific literature — to run large language models at less than 98% of the GPU utilization required by other leading models, then why have we been trying to invent GPU chips that do hard work faster, rather than just inventing new statistical methods that make the work itself easier?

It reminds me of the old tale of the consultant who realizes they make more money when they fail to solve the client’s problem quickly, because they’re charging by the hour. Our best-capitalized firms may lack incentives to create more efficient, competitive models. Perhaps leaders committed to scaling faster because they believed it would produce a first-mover's advantage. But that, too, has been revealed as false — the upstart firm was late to the game.

Shifting Priorities

The energy required to run these GPUs at large data centers is massive, and that energy has to come from somewhere. As states add new energy-generating capacity to support growing demand, tax- and rate-payers risk getting stuck holding the bag if we overbuild capacity for a technology that just became staggeringly more efficient.

Moreover, the environmental impact of such energy use is significant: studies suggest that training a large language model can emit as much CO2 into the atmosphere as a few thousand airplane trips — a non-trivial amount of pollution.
Why have we prioritized greater, more pollutive industrial scale, before solving the low-hanging mathematical efficiency problems that would reduce energy risk and environmental impact?

A Better Approach to Dominance in AI

The good news is that there is a path forward. I believe DeepSeek could prompt the industry to refocus its efforts on investing in intellectual property, true methodological innovation, unique protected assets like curated data and domain-specific applications.

The wake-up call from DeepSeek’s emergence isn’t just about global competition — it’s about the need for a fundamental shift in how we approach artificial intelligence. If the U.S. wants to maintain its leadership, it must move beyond brute-force scaling and capital-intensive expansion. The focus should be on smarter, more efficient innovation: refining mathematical methods, optimizing resource use and fostering an environment that rewards true breakthroughs over sheer size.

The next era of AI dominance won’t be won by those who spend the most — it will be won by those who think the smartest. DeepSeek has shown us that agility and ingenuity can outpace deep pockets. Now, it’s time for the U.S. to prove that it can still lead by doing what it does best — innovating with purpose.