Use AI bots to reply to followers

This insight is brought to you by… Modash

Marketers painted a very clear picture of Black Friday, saying they’ve missed out on creators because prices were too high. 

But when Modash surveyed content creators about holiday campaigns, they told a different story.

The (free!) report shows how increased demand, limited availability, and the opportunity cost of saying yes to one brand shape BFCM pricing - and what marketers can do differently.

Want to sponsor Science Says? Here’s all you need to know.

📝 Context

Topic: Social Media | AI
For: B2C. Can be tested for B2B
Research date: May 2026
Universities: University of Virginia, University of Illinois Urbana-Champaign

You’re trying to grow on Instagram and have seen some recent traction under your posts. People are leaving comments, and it’s getting harder to respond to everyone personally.

You’re tempted to use an AI agent to help you but worry it might backfire and hurt engagement.

Science says it’s not as bad as you think. Overall it will help you, but only if you do it right.

P.S.: A simple way to boost social media engagement is simply mentioning a rival in your content. These kinds of posts received up to 197% more likes on X.

📈 Recommendation

Use AI to help you interact with customers (e.g. comments, DMs). Make sure replies are relevant, sound like you and that you disclose it’s your AI (not you personally).

People will feel acknowledged and engage with you more over time.

🎓 Findings

• Using an AI bot to interact with followers (e.g. replying to comments) increases engagement on social media.

• Researchers ran an analysis of 22k followers’ interactions on 448 posts, and found that when people received a reply from an AI bot (a clearly “AI” labeled bot acting on behalf of the account) to their comment (vs those who did not receive a reply), they:

  • Commented 41% more often under later posts

  • Reposted the influencer’s content 43% more often

• The effect is:

  • Stronger when the community is engaged and the  AI reply was relevant, within the first 45 minutes of the comment, and matched the creator’s tone of voice.

  • Weaker when the creator already replies often and when the accounts offer paywalled content (e.g. paid Substack)

🧠 Why it works

• When we get a reply to a comment, we feel the creator is paying attention to us.

• This works even if the reply is from a bot (whereby people know it’s not the creator but AI disguised as the creator), as long as the response is timely and relevant.

• This feeling of being seen makes us more likely to engage.

Want to advertise to 30,000+ Science Says readers here? Click here.

✋ Limitations

• The study looked at one type of social media platform (Weibo) in China. The effect might differ across countries (e.g. Chinese users may be more accepting of AI) and platforms.

• Responses came from a personalized AI bot. People knew it was AI, not the creator. It’s unclear how people would react if they suspected the response was from AI, but it looked as if it was coming from the creator directly.

• Some messages might require a personal answer (e.g. emotional messages, personal ones) but the study didn’t test how the effect changes depending on message type.

• As AI technology evolves and people’s attitudes about AI change, the effect might too.

• We don’t know the difference between bot replies compared to human replies. Human replies could lead to more engagement, but this wasn’t compared.

• We also don’t know the effect of you commenting on other people’s accounts using AI bots. It likely backfires.

• Few platforms make it easy to use such bots, some ban them outright (e.g. Linkedin). However, the effect is still relevant for some major platforms that do allow for these kinds of integrations (e.g. Instagram with Manychat).

• The study effect might not be as strong on branded accounts without an audience-facing person.

👀 Real-life example

Many creators and brands receive comments but don’t respond to them.

❌ Issue: This post got a few comments with relevant questions, but no replies. This is likely hurting engagement on future posts.

✅ Solution: These are a few simple things creators could try out:

• Use AI to help with replying to comments. For example, they could train it on their voice and own content and use it to draft responses.

• Make sure that when the AI bot replies, it clearly says it is an AI, and not the creator (e.g. “AI bot of [Creator]:”)

• Respond with clever humor. Funny but still relevant responses make people even more likely to engage.

• Show empathy to those leaving complaints and confused comments (as in the example above) by leaving a thoughtful reply, or even suggesting a way to get into direct contact.

🔍 Study type

Market observation (of 22,645 follower comments on Weibo between July 1 and August 31 in 2024).

📖 Research

When Influencers Delegate Replies: How Social AI Agents Shape User Engagement. Information Systems Research (May 2026).

🏫 Researchers

• Maggie Mengqing Zhang. McIntire School of Commerce, University of Virginia

• Yang Gao. Gies College of Business, University of Illinois Urbana-Champaign

• Jingjing Li. McIntire School of Commerce, University of Virginia

• Steven L. Johnson. McIntire School of Commerce, University of Virginia

Remember: This is a new scientific discovery. In the future it will probably be better understood and could even be proven wrong (that’s how science works). It may also not be generalizable to your situation. If it’s a risky change, always test it on a small scale before rolling it out widely.

🎁 Bonus mini-insight

Read the full insight here if you are already a member.

🎓 Insight: What makes people like video ads

📈 Recommendation: Create content with these elements to make people more likely to engage with it:

• Highly emotional at the start but rational at the end (e.g. providing a solution).

• Evoke empathy throughout (e.g. narrating how people feel).

• Have dialogues (both textual and voice-over).

• Engage both sight (e.g. through zoom-ins and outs, dynamic images) and hearing (e.g. sound, music).

✋ Careful: It’s unclear whether the emotions evoked should be positive or negative, and how the effect would translate to very short content where the narration is more compressed

Here is how else Science Says can help your marketing:

• 📈 Join the Science Says Platform to unlock all 300+ insights, real-world case studies, and exclusive playbooks

• 📘 Boost your sales and profits with topic-specific Science-based Playbooks (e.g. Pricing, Ecommerce, SaaS, AI Best Practices)

• 🔬 Get science-based thought leadership content for your brand. We help market leaders in AI, tech, and across B2B enterprises, build authoritative thought leadership using a scientific perspective. Find out more.

🎓 It took me, a PhD analyst, and a marketing pro 15 hours to accurately turn this 20-page research paper into this 3min insight. 

If you enjoyed it please share it with a friend, or share it on LinkedIn and tag me (Thomas McKinlay), I’d love to engage and amplify! 

If this was forwarded by a friend you can subscribe below for $0 👇