AI Briefing: How tech companies are talking about AI on social media
From sponsoring the Olympics and major film festivals to buying ads that run on TVs and in taxis, AI providers have spent the past year looking for ways to break through the noise. But what does the volume of conversation look like for the masses on social media?
To get a sense of how often tech brands have mentioned AI on various social platforms, Digiday asked Sprinklr, a customer experience management platform, to analyze paid and organic content posted in 2024. To narrow the focus, Sprinklr searched for companies’ social media posts for mentions of more than a dozen AI-related terms and hashtags: “AI,” “Artificial Intelligence,” “Generative AI,” “Gen AI,” “AI Generated,” “GPT Powered,” “GPT-Powered,” #ai, #artificialintelligence, #generativeai, #genai, #aigenerated or #gptpowered.
From flagship AI products and companies to words related to AI’s promises, the volume of conversations provided at least some sense of how tech firms are describing their offerings across branded social accounts. It also shows which AI players are being talked about the most — both by the companies themselves and social media users that mention them.
Sprinklr also looked at a content few companies with major AI platforms — Google, Microsoft, OpenAI and Adobe — to analyze mentions of their flagships products alongside the AI-related terms. According to the data, the most mentioned company was OpenAI, which was cited 123,000 times in posts from more than 100 tech firms. Following OpenAI were Google with 85,000 mentions, Apple with 43,000 and Microsoft with 59,000.
Sprinklr also analyzed total global user mentions of AI products and found ChatGPT was the most common. It came up in more than 4.3 million AI-related posts, followed by Gemini (475,000), Copilot (144,000) and Firefly (57,000).
More than 5,000 of Google’s posts mentioned Gemini across social accounts for Google, Google Pixel and Google Cloud. Meanwhile, Microsoft’s main account and a separate one for Microsoft Copilot mentioned Copilot about 300 times. Other tech companies with a lot of AI mentions so far this year include Salesforce, the Israeli customer services firm NICE, AWS, Intel and Oracle.
The analysis doesn’t show the entire scope of AI-related social conversation. For example, it doesn’t include every major AI product brand and its accompanying social channels for even some well-known players such as Amazon Q, Meta AI and IBM Watsonx. However, it still illustrates how tech companies and people talk about AI across various platforms.
“A lot of common words in the [in the analysis] were functional and focused on actions: ‘create,’ ‘learn,’ ‘build,’” said Brandon Boatwright, an assistant professor at Clemson University and director of Clemson’s Social Media Listening Center. “There’s this arms race for these tools to be integrated.”
It’s also worth looking into the sentiment of the social content, according to Boatwright. For example, companies use positive words like “excited” and “amazing” to describe AI, but what about all the people online talking about it? That could help better understand how people feel about AI and help companies respond to various areas of confusion or concern.
Despite ongoing battles with advertisers, X is still driving plenty of reach. OpenAI had 80.32 million earned impressions while Google got 58 million and both Microsoft and Adobe had around 3 million. Across all Twitter, Meta and Instagram platforms, links were the most engaged type of content, followed by photos, video and text. OpenAI’s content had the highest level of average engagement (3,100 engagements per post), followed by Google (1,000), Microsoft (600) and Adobe (215).
There’s both a “top-down and bottom-up” approach with marketing AI to various audiences, said Tom Lyons, president SociStudio, a DNY-owned social agency. AI providers pitch complex solutions for enterprise adoption, while many of the same tech companies are also pitching various consumer-facing tools. Although AI is still a “toe dipper” for many, he thinks it’ll be especially key for AI players to persuade small businesses to try new tools for tasks like creating content. But unlike some past emerging tech trends, AI is easier for the average person to understand its output.
“[Web3] was a bit of a head-scratcher for a lot of people,” Lyons said. “But if you go into MetaAI, you are not scratching your head. You understand instantly what just happened and you essentially understand you just went from first base to third base in whatever you’re trying to write or create or find …[Generative AI] is more akin to Google when Google first launched than it is to Web3.”
Although enterprise companies in recent years have sought to create more consumer-level marketing to reach target audiences, the sophisticated nature of AI might mean developer audiences need to understand it first. And while tech companies talk about AI’s potential, there’s still the risk of over-promising and under-delivering.
AI companies’ range of target audiences also requires more targeted and differentiated messages, according to Matt Wurst, chief marketing officer of Genuin. That means simplifying the tech for some audiences but also going deep enough for technical crowds. Wurst also noted that companies need to be talking more about what people can do with the tech rather than what companies have created themselves.
For giants like Google and Microsoft, using parallel messaging in other categories has become common, Wurst said. With Gemini and Copilot being rolled out for enterprise and consumer products, the location of each message is key.
“Personalization and contextual relevance isn’t something you need to be putting on billboards or in mass media,” said Wurst. “At first it’s about translation, and then it’s about interpreting … This is absolutely one of those cases where the messaging is not always going to be right early on, because the products are more sophisticated than I think anyone even realizes.”
Prompts and Products: AI news and announcements
- The FTC announced new rules that bans fake reviews, including fake AI-generated reviews.
- Google is expanding its AI Overviews feature for AI-powered search and plans to provide citations.
- Anthropic debuted a new feature called Prompt Caching, which promises to improve speed and decrease costs for API calls.
- Snapchat rolled out new tools for advertisers including a new way to generate ad copy using AI.
- SAG-AFTRA and the AI voice startup Narrativ announced a new deal to let actors work with Narrativ to have a digital replica of their voice used in brands’ audio ads, and be compensated for it.
- Meta replaced its Crowdtangle tool with its Meta Content Library, despite calls for the company to leave it in place until after this fall’s election.
- Hubspot added a new “AI search grader” to help brands test how they show up in various AI search answers.
- XAI released Grok 2, which includes an image generation feature for X. The update follows a letter several secretaries of state sent to X owner Elon Musk asking him to address the spread of election misinformation on the platform.
- A new survey from Digitas and Vox looks at the correlations between AI and creativity.
- The startup BattlegroundAI expanded to beta access for political advertisers that want to create AI-generated content for down-ballot races.
Other AI stories from across Digiday
More in Media
A look at the publisher quandary over ad curation
At a Digiday Summit, publishers confront the fine line between revenue and oversight.
Creators weigh content decisions and costs of election-driven marketing blackout
Agencies, creators and brands respond to the election-driven blackout on marketing.
Media Briefing: European publishers sound off on site traffic struggles
This week’s Media Briefing looks at what publishers attending Digiday Publishing Summit Europe had to say about site traffic challenges and opportunities to better monetize page visits.