Are you referring to? What happenns when demand falls? or why the demand fell? or something else.
I am referring to where you wrote,
There is zero justification for the second sentence. It doesnât appear in the original report. It appears to be entirely fabricated by you.
The simple fact is, there are no year on year metrics for engagement with AI to look at, this is currently âYear 0â Interpersonal communication and sharing sites do not suffer reductions in growth over holiday periods as people are posting about their holidays. Sites that have a distinct educational or work focus do.
It should also be noted that while ChatGPT is the most user friendly way to interact with AI currently, it is still rather text based and specialised for those with a proclivity for high technology. What those usage metrics do not describe is the explosion in 3rd party applications based on the API which I believe increased by 3.somthing%
There will be providers of less moderated public facing AI portals, and they will be popular for users that wish to take advantage of them, however, there is usually a separator between providers of âadultâ services and those who do not, now the debate as to why that is as it is and the social reasons for it are a topic worthy of debate, but probably not on an AI developer forum.
Here is the article I read. Censorship is a restriction to me:
arstechnica .com/tech-policy/2023/07/chatgpts-user-base-shrank-after-openai-censored-harmful-responses/
Take the space out between arstechnica and .com. It will not let me post the actual link.
Censorship is certainly a restriction, but the article you appear to be relying on has taken two separate source articles :
- https://www.similarweb.com/blog/insights/ai-news/chatgpt-traffic-drops/
- https://www.nytimes.com/2023/07/02/technology/ai-chatbots-misinformation-free-speech.html
And implied a possible connection between them (emphasis added),
These guardrails could be blocking chatbots from providing legitimate responses, a New York Times report this week suggested, and, thus, sending frustrated users away from Big Tech tools like ChatGPT and toward open source, uncensored chatbots that are increasingly becoming available online.
Which you then stated much more definitively,
So, we have one article taking about a dip in traffic with no causal suggestions made. Another article taking about how there is a recent deluge of uncensored models but no suggestion that people are abandoning Big Tech models. A third article which essentially said, âmaybe the cause of the dip is people moving to uncensored models.â Which you then reported as âGPT has too many restrictions so people have started going elsewhere.â
Do you now see the issue I tried to address in your first post?
Now, letâs look at something else,
This is the visit data for https://www.wolframalpha.com, notice anything?
Traffic to WolframAlpha dropped by 13.9% May-to-JuneâŚ
Is that also because of ChatGPT censorship?
Here is the data for https://www.desmos.com
Traffic dropped 34% May-to-June.
Hereâs the data for https://www.wikipedia.org.
Traffic dropped by 6.4% May-to-June.
Are you seeing a pattern yet?
While there is, no doubt, some number of people who have abandoned ChatGPT in favor of less-restricted models, the far simpler explanation is that sites used extensively (though not exclusively) by students typically experience a substantial decrease in traffic May-to-June.
ChatGPT, rather famously, has been used extensively by students.
1 Like
That is the beauty of the internet, if you look around long enough you can find different data and conclusions.
I only read one article and the gist I was trying to relay was that if GPT-4 API was slow in comming that the lower level of resource demand would help get GA faster. I didnât spend hours researching different interpretations of the data from different data sources. Due to the title of the article I read my statement about the reason seemed more heavily weighted.
I wasnât making any commentary on big tech, just pointing out that the earliest innovators usually are not the long term winners. My Space, Informix, HP/CPQ, IBM, Sun, etcâŚ