ChatGPT versus metaverse platforms google trends

In contrast to the "The Metaverse is dead, long live generative AI!" narrative you might have heard recently, here's some data strongly suggesting otherwise: On Google Trends, ChatGPT for all its recent buzz has never eclipsed general interest in leading metaverse platforms Minecraft and Roblox. (See above.)

Even more notable, ChatGPT has been consistently trending downward on Google Trends since April.

According to François Chollet, who works in Deep Learning at Google, the reason for that is obvious:


https://platform.twitter.com/widgets.js

"'Producing homework' remains the #1 application of LLMs'," as she puts it bluntly. "Which is remarkable, because the point of homework is to do it yourself — the actual deliverable has zero value to anyone."

So as high school and college students started to go on break in May, the need for a program like ChatGPT to cheat on homework — I'm sorry, to serve as an informational resource for report preparation — rapidly goes away. And indeed, general traffic to ChatGPT site OpenAI also started trending down in May, according to SimilarWeb (see above).

I don't doubt that's part of the reason why ChatGPT is trending downward, as compared to Roblox and Minecraft, but another factor is key: Metaverse platforms are dynamic online communities, while an LLM like ChatGPT is mainly an informational resource. You go there for an answer, you get one (or several) and then you leave. 

You can see that in traffic trends of a metaverse platform like Roblox, as compared to ChatGPT:

While OpenAI's site gets about twice the raw traffic as Roblox.com (according to SimilarWeb), actual engagement is much thinner, with a bounce rate of 42% and an average visit duration of 4 minutes, 25 seconds. (Roblox's bounce rate is under 19%, with an average visit duration of 13 minutes, 38 seconds.) 

This also fits the data I noted in May, that leading metaverse platforms still have more unique users than leading generative AI platforms.

Or to put all this another way: an LLM like ChatGPT might give you useful answers (sometimes), but it doesn't create nor sustain a community around it. 

Unfortunately, the community of marketers and evangelists who still insist on saying, "The Metaverse is dead, long live generative AI!" will be around a bit longer. At least until the next buzzy concept they don't necessarily understand comes along.

amzn_assoc_tracking_id = "newwornot-20";
amzn_assoc_ad_mode = "manual";
amzn_assoc_ad_type = "smart";
amzn_assoc_marketplace = "amazon";
amzn_assoc_region = "US";
amzn_assoc_design = "enhanced_links";
amzn_assoc_asins = "1394155816";
amzn_assoc_placement = "adunit";
amzn_assoc_linkid = "09778d9a32dae1475f81139abdba7c70";

//z-na.amazon-adsystem.com/widgets/onejs?MarketPlace=US

Posted in ,

2 responses to “ChatGPT Outpaced By Leading Metaverse Platforms on Google Trends as Visits to OpenAI Website Trend Down”

  1. Iggy 1.0 Avatar
    Iggy 1.0

    Two thirds of our student respondents are using ChatGPT or similar AI.
    There’s a use-case there, ethical or not, that cannot be made for VWs or VR rigs. I think you make a mistake to lump these technologies together; using AI is not about community, no more than using a search engine.
    You want that for college students? Look at their use of social-media, esp. Tik Tok.
    After the hype and unfounded Doomerism dies down, AI is going to be a big part of our lives, albeit a largely invisible one as a plug-in for Word, Grammarly, and search engines.
    I still don’t see metaverses as more than something for a niche community. I drank that Kool-Aid back in 2007 or so and will wait a bit more before taking another sip. The Metaverse is not dead. It’s just one island in a big ocean for now.

    Like

  2. Nadeja Avatar
    Nadeja

    “an LLM like ChatGPT is mainly an informational resource”.
    Large Language Models (LLMs) such as GPT-4 (ChatGPT is the application, not the model) can assist you with various tasks related to text processing; but please note that they they are NOT reliable sources of information! It’s dangerous to use them that way: they remember their training data, more or less, but the data may be outdated and they make mistakes or hallucinate information that does not exist. A recent case of a lawyer who used LLM as info resource and got into trouble illustrates this danger.
    You should not rely on LLMs as info resources, but rather as text processing assistants: text and grammar improvement, summarizing, idea generation, etc. (students don’t only use them to cheat). They help with coding and other tasks too.
    If you really want to use a LLM to find information, use at least one like Bing Chat in Precise mode (that hallucinates very little; Creative mode instead is meant to invent things, poems etc), which uses the search engine to search for updated information and summarizes the results. I’d verify anyway.

    Like

Leave a comment