The chatbot arms race is a popularity contest. Good performance might take a model around the block, but a story that captures popular imagination will take it around the world.
New research released this week found that the top large language models (LLMs) based on social media mentions between October 1, 2023, to March 31, 2024, were ChatGPT, Gemini, and Claude.
This is according to data collected by GlobalData’s Social Media Analytics Platform, which tracks more than 250 million posts across X (formerly Twitter) and Reddit, including 10,000 “influential” Twitter accounts and 3,000 Reddit channels.
ChatGPT had 69.8% of the share of the mentions, followed by Gemini with 14%, and Claude with 5.3%.
The runners-up were Grok with 3.4%, Llama with 2.9% (includes Llama 2), Mixtral with 2.9%, PaLM with 0.6%, LaMDA with 0.5%, Falcon with 0.4%, and Qwen with 0.3%.
The top three stood out not just because of their performance but also because they captured the community’s imagination.
Key Takeaways
- ChatGPT, Gemini, and Claude dominated social media mentions between October 2023 and March 2024.
- ChatGPT led with 69.8% of mentions, followed by Gemini at 14% and Claude at 5.3%.
- ChatGPT’s first-mover advantage and continuous updates, including the release of GPT-4 Turbo, likely contributed to its prominence.
- Gemini’s integration with Apple’s technology and Claude’s safety-focused AI assistant status also propelled their mentions.
- However, emerging models like Llama 3, with open-source accessibility and competitive performance benchmarks, could disrupt the top three’s position.
- Differentiating through different narratives or values, alongside performance, remains crucial for AI vendors to capture people’s interest.
Mention | Share |
ChatGPT | 69.8% |
Gemini | 14.0% |
Claude | 5.3% |
Grok | 3.4% |
Llama | 2.9% |
Mixtral | 2.9% |
PaLM | 0.6% |
LaMDA | 0.5% |
Falcon | 0.4% |
Qwen | 0.3% |
The Story Behind the Top 3 LLMs
With the generative AI market being as competitive as it is, a model not only needs to offer solid performance, but it needs to be differentiated from other competing tools.
Smitarani Tripathy, a social media analyst at GlobalData, explained to Techopedia that the report “underscores the growing impact and integration of LLMs in various sectors, reflecting their widespread acceptance and the public’s fascination with AI technologies.”
When asked why ChatGPT, Gemini, and Claude were the top three mentioned tools, Tripathy suggested that ChatGPT had a first-mover advantage and likely ranks high on social media due to the release of continuous updates — GPT-4 Turbo being one of the most prominent.
The report notes that social media discussions around ChatGPT spiked on November 6, 2023, when OpenAI unveiled GPT-4 Turbo at its first-ever Dev Day developer conference.
Likewise, decisions to broaden the application programming interface (API), reduce costs, and introduce multimodal capabilities also increased interest.
With regards to Gemini, Tripathy suggests that Gemini “might be propelled” by its integration with Apple’s technology and the potential to work alongside Siri.
Finally, Claude “may owe its place” to the release of Claude 3 Opus in March 2024 and the fact that it outperformed GPT-4 for the first time in Chatbot Arena.
At the same time, each of these top three has a simple story to tell; ChatGPT ignited the LLM arms race in November 2022, Gemini brought to market a multimodal family of models, and Claude introduced a safety-focused AI assistant.
Stephen Fraga, CEO at Prompt Yes!, offered Techopedia a similar but ultimately different take on why the top 3 LLMs sat so far above the competition.
He said:
“ChatGPT was the first LLM that was shockingly impressive and fully available to the public; now it has a recognizable brand name.
“Meanwhile, Gemini piggybacks on Google Search’s ubiquity, so everyone is exposed to it instead of having to go to a specific website like ChatGPT.
“And Claude improved on one of the main drawbacks to ChatGPT — too short a context window; this was especially helpful to programmers that needed their LLM to keep track of long chunks of code.”
So for Fraga, the top stories surrounding each product were that ChatGPT had the first mover advantage, Gemini had a Google Search connection, and Claude could support larger inputs.
Binny Gill, CEO at Kognitos, noted he was surprised by the top three:
“It’s interesting that Gemini holds a spot in the top three after taking its licks for its DEI debacle on the very platforms where people are apparently using it! This is likely due to its multi-modal strengths with video and images that play well in social media settings.”
Gill is referring to the PR storm that Gemini recently faced for being “woke” after a viral post emerged showing the model generating an image of the US founding fathers, which incorrectly included a black man, and another example emerged depicting a black man and an Asian woman as a representation of German soldiers from World War Two.
How Will the Top LLMs Change?
As new models emerge, the top three LLMs may be displaced. At the moment, one of the most promising contenders to do this is Llama 3, a family of multimodal models that dropped last week.
Llama 3 is a wildcard now because it’s one of the most powerful open-source models available that can be accessed for free via the Meta AI Assistant, and it outperforms top proprietary solutions like Gemini Pro 1.5 and Claude 3 Sonnet on key performance benchmarks.
Llama 3’s biggest asset is its connection to open-source: At least on paper, it represents democratized AI development and a welcome alternative to opaque and unaccountable blackbox AI. And there’s potential for more Llama-based LLMs to follow suit.
Gill added:
“I expect to see a complete explosion of LLMs, with many emanating from open source. There is a tremendous opportunity for new LLMs to differentiate themselves through specialization, either in subject matter or in perspective (including bias).”
Elon Musk’s witty conversational chatbot Grok and Perplexity AI’s AI-assisted search assistant also present potential wildcards to the market, though Fraga argues that Grok’s paywalled nature may hold its adoption back.
The Bottom Line
LLMs that capture user interest the most on social media have a story to tell. So AI vendors that want to increase adoption not only have to offer cutting-edge performance, they also need to differentiate their story or values from other competitors.
As we are still in the early days of AI, there is plenty to play for.