TikTok Testing AI-Generated “Virtual Influencers” for Sales & Promotion

MIT Researchers Develop Technique to Safety-Check Chatbots

Read time: 3 minutes

Hi AI Pro!

Marketing pros and their tech colleagues continue the dance with evolving AI tools and practices, as they discover what works and what doesn’t—and how they feel about it all…


  • Top Story: TikTok Testing AI-Generated “Virtual Influencers” for Sales & Promotion

  • Damage Control: MIT Researchers Develop Technique to Safety-Check Chatbots

  • Market Trends: Research: WhatsApp Still Reigns Supreme in Conversational Marketing

  • Early Money: Data Scraper for Training LLMs Apify Raises $3 Million

  • Job Satisfaction: MRII Study: Majority of Market Researchers Are Positive Towards AI

  • Today’s Master Marketing Tip: Marc Benioff, chair, CEO, and co-founder, Salesforce


TikTok Testing AI-Generated “Virtual Influencers” for Sales & Promotion

As a significant share of marketing campaign spend goes to engage online influencers, TikTok is currently testing an AI-powered feature that generates avatars instead of real humans for selling products and services on its social media platform.

TikTok’s AI avatar engine, which is not generally available yet, creates virtual characters that read scripts based on prompts input by advertisers, and so would assumedly be able to take market share from real-life human influencers.

According to a report by The Information, inside sources at TikTok say that tests of the AI avatars have been less-than-stellar so far, producing fewer sales than human influencers. The company reportedly doesn’t consider the new avatar feature ready for release, according to the article.


MIT Researchers Develop Technique to Safety-Check Chatbots

The rise of AI chatbots that generate human-like responses, assist with various tasks, and provide useful insights, has led to some concerns about their potential to generate inaccurate and/or harmful content.

In response, researchers at the Improbable AI Lab at MIT and the MIT-IBM Watson AI Lab have developed a new approach to improve the “red-teaming” process for chatbots, using machine-learning strategies to test for vulnerabilities.

This new “curiosity driven” approach developed at MIT involves training a separate large language model that automatically generates prompts in an attempt to trigger undesirable responses from the chatbot tested.

“They do this by teaching the red-team model to be curious when it writes prompts, and to focus on novel prompts that evoke toxic responses from the target model,” according to a recent report in MIT News.


Research: WhatsApp Still Reigns Supreme in Conversational Marketing

An analysis of 473 billion digital conversations during 2023, performed to identify the latest global business messaging trends, reveals that conversational customer experiences and generative AI tech “have created a long-lasting impact for how people and brands interact.”

According to Infobip, a cloud communications platform that conducted the study, “As consumers look for a deeper connection with the brands they buy from, conversational messaging channels are seeing the fastest growth.”

Infobip reported a 137% increase in mobile app messages in 2023 through its platform compared to 2022, a 73% rise in social media messages, and a 63% increase in chat app messages.

“WhatsApp remains the top digital channel for conversational marketing, driven by new features that enable customers to start and complete a purchase in a single WhatsApp chat window,” according to Infobip.

The company also reports seeing significant increases in other messaging apps such as Telegram, Line, Viber, and Messenger, with data showing a 29% increase in mobile app messaging for marketing in 2023 compared to the previous year.

WhatsApp also continues to stand out as the primary channel that brands use for conversational support, with businesses sending 90% of support messages over the app.


Data Scraper for Training LLMs Apify Raises $3 Million

Apify, a Prague-based developer of web data scraping technology that’s used for training large language models (LLMs) that drive AI-generated products, among other types of applications, has raised $3 million in a new round of venture capital.

Founded in 2015 through the Y Combinator Fellowship program in Mountain View, Calif., Apify allows users to extract text and images from websites to generate training datasets for new AI models. The company’s technology also supports extraction of documents from the web and loading them into vector databases for querying and prompt generation.

“We have been supporting Apify for eight years, and we have been sure of its potential every step of the way,“ says Ondřej Fryc with Reflex Capital, the company’s lead investor.

“[Apify’s founders] have already built a global business with monthly recurring revenue in the high hundreds of thousands of dollars that are now rocketing. We see the investment as an opportunity to further support this growth,” Fryc said.


MRII Study: Majority of Market Researchers Are Positive Towards AI

A new online survey of 369 market research professionals interviewed between February 22 and March 27, 2024, by the Market Research Institute International indicates overall optimism about the role that AI can play in their daily work life.

Among the key findings:

One-third currently use generative AI, jumping up 25% from the previous year's MRII study.

Nearly 60% believe that both generative AI and AI/machine learning will be some of the highest impact skills of the future, for which training is needed.

74% percent feel favorable about AI as it relates to their job and company, with 72% believing AI will make their job better.

Most are not worried about losing their jobs to AI: 60% show no concern; 37% have "a little" concern, few are very concerned.

Consultants and agency-side researchers are more likely to use AI in their work (47%), while corporate researchers have been slower to adopt the technology, with 25% reporting current usage.

"The results are fascinating and can serve as a guide to leaders in our industry about the ways their workforce is thinking about AI, as well as for any researcher who wants to benchmark their views against peers,” said Ed Keller, executive director of MRII, in a press release.


“Artificial intelligence and generative AI may be the most important technology of any lifetime.”

— Marc Benioff, chair, CEO, and co-founder, Salesforce


Want to learn how to get your brand referred at scale to your target customer? Check out Presspool.ai to learn more.

Let’s win together!


How would you rate today's newsletter?

Login or Subscribe to participate in polls.