Generative AI, meet social media

The Future. Major social media platforms like Meta and Snap are unveiling their own Open-AI-based chatbots and language models. The two offerings have a variety of use cases, but it’s not yet clear if either will manage to make money– or avert ChatGPT’s misinformation disasters. If they fail, gen AI may struggle to become a big income stream for brands.

Visions of the future
Meta and Snap’s new generative AI products differ widely in their implementation and prospects.

• Snap has released its new chatbot, “My AI,” powered by ChatGPT and available to its 2.5 million paying Snapchat Plus users. My AI is designed to write haikus, recipes, trip itineraries, answers to trivia questions, and other recreational or artistic content.

• Snap will store all My AI interactions for research and development. The firm cautioned against telling secrets to My AI or relying on it for advice, since ChatGPT’s misinformation issues could still occur.

• Meanwhile, Meta unveiled its new language model, LLaMA (Large Language Model Meta AI), a vast collection of language models that vary in size.

• Meta will make LLaMA’s collection of language models available on a case-by-case basis to government, civil society, and academic research organizations. Its business applications aren’t clear yet.

Pandora’s box
There’s a huge risk that a company’s concern for their profit margin could motivate them to unleash their gen AI technology in irresponsible or downright unscrupulous ways. Compared to OpenAI, though, LLaMA’s smaller language models are apparently easier to debug and keep from spreading misinformation.

The answer probably depends on whether and how either platform will monetize these programs; so far, businesses don’t seem particularly excited about LLaMA, and Snap’s My AI seems more like a novelty draw than a branding opportunity. But someone will try soon enough.