Article's Content
Article's Content
It’s rare that the world gets to see a truly disruptive technology develop in front of their eyes in real time — but that’s what we have with AI.
We’re all getting to experience breakthroughs that experts are putting on par with the Manhattan Project and cracking the genome.
For many, it’s probably already fallen into the background (it has for myself, more than a few times). But when you stop to think about the amount of time, money, and resources going into the incremental advancements, the small hops and the large bounds forward, it really is amazing.
It seems like just a few months ago, OpenAI’s ChatGPT first went viral. And yet:
GPT-3 is now ChatGPT o1.
Claude is now Sonnet 3.5
Bard is now Gemini.
Not to mention the countless open-source models that are also pushing the industry forward. And it’s that concept — open source — that I want to talk about today. Because the research that is being done to fuel the AI arms race is becoming an incredibly important resource in and of itself.
The AI Research Landscape: How Leading Companies Stack Up
According to Stanford’s 2024 Artificial Intelligence Index Report, the total number of published articles tripled from 2010 to 2022, exploding from 88,000 to 240,000. Countless hours of research were necessary to bring the predictive and generative AI models we use today to life.
Not surprisingly, the usual Big Tech suspects — Microsoft, Google, Meta, and Apple — were heavily involved. In fact, a group of Google employees worked on the infamous Attention is all you need study outlining the transformer technology that forms the foundation of LLMs.
Since then, AI-vy leaguers have become one of the most important, sought-after talent pools in the world, with the Big Tech companies and AI startups scrambling to keep the best and brightest.
Why? Because better minds mean better breakthroughs and better products.
That’s why it’s so important to look at the research coming from these companies. The company with the best research is also likely to be the company that produces the most useful tool.
With that in mind, I took a quick look at the body of research that five of the leaders in AI have listed on their websites: Microsoft, Meta, GoogleDeepMind, OpenAI, and Anthropic.
At first, it seems like Microsoft is running away with the race and leaving everyone else in the dust (considering its close ties to OpenAI, that may actually be true). They have more published AI research than the other four companies combined just in 2024 alone.
But quantity is only one side of the story.
What about the quality of the research?
And the impact on the industry?
It’s the same reason why Ross is always talking about content distribution. It’s great to publish all that content, but what’s the point if you aren’t getting it out there for people to read?
It just so happens that we’re not the only ones asking these questions. The 2023 study Who is leading in AI? found that, while Microsoft led the way in terms of total publications, it was actually OpenAI that led the way in terms of the number of citations per year per author and Google who led the way in terms of innovations that have been adopted into LLMs.
So, the race is a lot closer than it seems when you look at the quality of research and how it actually impacts the industry. That’s especially true considering companies like Anthropic have come on strong in the last few years, pioneering spaces like AI safety and alignment (not to mention my favorite model LLM, shout out Claude).
Taking our lens towards the marketing and product growth landscape, there’s actually another metric that you can use to see how impactful the AI research from these companies is: backlinks. The measure of authority in research is citations — for the rest of the web, it’s how many websites co-sign your work by linking to it.
With over 742,000 all-time backlinks, OpenAI is leading the way as the AI authority in the eyes of the general public. This makes sense, considering they were the catalyst for the boom back in 2022 and have the most popular consumer AI tool. Next comes Meta with over 37,000 all-time links, followed by Anthropic with 32,300, and Google DeepMind with 17,400.
Now let’s take a closer look at the type of authority these companies are building in the public eye.
Behind the Research — Where Leading AI Companies Are Building Authority
As one of the most popular debuts of a software product of all time, it’s easy to assume that all of these companies are chasing the ChatGPT trajectory and that OpenAI would continue to focus on building its lead.
But when we take a step back to look at the AI research focus behind Microsoft, Google, Meta, and Anthropic, we get a more comprehensive picture.
Microsoft
What’s interesting is that they’re not just focused on making AI smarter—they’re actually trying to make it work better with humans. You know those AI copilots that are popping up everywhere? That’s Microsoft’s thing. They’re betting big on the idea that AI works best when it’s helping doctors diagnose patients or helping teachers create better lesson plans, not trying to replace them.
And while everyone’s talking about AI chatbots, Microsoft is focusing on how AI can tackle some pretty heavy scientific problems. They’re letting AI loose on everything from understanding how proteins fold to figuring out new physics theories.
Backlink Leader (Last 5 Years):
- AI4Science to empower the fifth paradigm of scientific discovery: 22,052 backlinks
Like Microsoft, Google’s AI research is focused on using AI to solve complex challenges and advancing technologies that have a broad societal impact. Domains like health science, quantum computing, and the next generation of LLMs.
The company’s commitment to responsible AI ensures that their innovations positively impact users while addressing global challenges. Through Google Research and DeepMind, they continue to shape the future of AI by fostering collaboration and pushing the boundaries of what’s possible.
Backlink Leader (Last 5 Years):
- Human-level control through deep reinforcement learning: 2,326 backlinks
OpenAI
If there’s one company that’s been setting the pace in the AI race, it’s OpenAI. They burst onto the scene with ChatGPT, but that was just the beginning. They want to build an artificial general intelligence that can think at a human level — but without the sci-fi nightmare scenarios.
They’re taking a full-spectrum approach to research: language models that can actually understand what you want them to do, DALL·E turning imagination into images, and Whisper handling speech like it’s second nature. Their latest venture, Sora, is pushing into AI video generation, showing they’re not slowing down anytime soon.
Backlink Leader (Last 5 Years):
- GPT-4 Technical Report: 137,460 backlinks
Meta
Meta might be best known for Facebook, but their AI research team (FAIR) has actually been in the game for over a decade. While everyone else is chasing chatbots, they’re tackling some pretty interesting challenges that most people aren’t talking about.
Take their work on SAM 2, which is basically teaching AI to understand what it’s looking at in images with incredible precision. Or Video Seal, which is trying to solve the whole “Is this video real?” problem that keeps everyone up at night. They’ve even built something called DIGIT that lets robots actually feel what they’re touching — talk about future tech. Their latest project, MovieGen, is jumping into the AI video generation race.
Backlink Leader (Last 5 Years):
Anthropic
Anthropic’s approach to AI is a bit different from the pack. While everyone else seems to be racing to build bigger and flashier models, Anthropic is focused on a deceptively simple question: How can we make AI that actually does what users want it to do?
The Anthropic research team is digging into some pretty fascinating areas — trying to understand what’s actually happening inside these AI models, and making sure AI stays helpful rather than going off the rails. It’s not just theoretical either — they’re developing real tools to measure AI capabilities and spot potential problems before they happen.
What makes them interesting is their systematic approach. They’re big on finding actual patterns in how AI scales and behaves, rather than just hoping for the best. Their research papers tackle everything from how AI models “think” to how they might try to trick us — real “trust but verify” stuff. And unlike some companies keeping their cards close to their chest, they’re pretty open about sharing what they learn.
Backlink Leader (Last 5 Years):
- Alignment faking in large language models: 12,747 backlinks
Considering Original Research as a Method for Building Authority (and Links)
For those of us who aren’t working on the cutting edge of AI research, there are still some lessons to take away from all of this.
First, it’s a great reminder once again that quality does not equal quantity. You can publish an ocean of research, but unless people are reading it, it’s not likely to do the world much good. (So, you know, go distribute that prized blog post again.)
Second, while the AI industry might be suffering from an abundance of high-quality research from the best and brightest, the same may not be true of your industry. So consider this another endorsement to invest in that big-lift research piece you’ve been thinking about.
Not only is it a great opportunity to push your industry forward and establish your brand as a thought leader, but it’s also great for building authority and backlinks. A little while ago, Ross talked about how tools were the cream of the crop for earning backlinks and building site authority.
But look at number four on that list.
What type of content is MOST LIKELY to get links?
We wanted to find out.
So we took a look at some of the top Martech websites in the world and analyzed their most frequently linked to pages to uncover what is the most linkable SEO content.
The answer was tools. pic.twitter.com/sfHinISlBz
— Ross Simmonds (@TheCoolestCool) February 4, 2023
Original research has always been a vital yet underappreciated part of the marketing toolkit. Which is why we always prioritize it at Foundation and recommend it for our clients. It may not be as sexy as a brand overhaul or social campaign, but high-quality research is equally, if not more, important for establishing authority in an industry.
Want to know more about how Foundation can help you create high-value, highly shareable research? Get in touch.
