Hello!
Thank you ??
I have no plans to implement OpenAI’s services into TSF — I built bespoke generators, which are fast, becoming more accurate with every update, and are practically free to operate.
Here are the issues with implementing GPT:
- Privacy and Security concerns: If we’d implement GPT-based software, we’d have to read your content and offload it to a third-party data processor. That content may be intended for a private audience, contain location information, and other Potential Person Identifying information that would be stored indefinitely, and that content may be reshared by accident to another user.
- Loss of control: The moment we implement such a service, we’re bound to abide by still-emerging legislation. This can backfire when we must close this service for any reason whatsoever.
- Power consumption: I made TSF fast and green — TSF generates a description under a millisecond on a single-core 15-watt CPU. GPT AI is slow and polluting — ChatGPT generates a description in about 10 seconds using multiple 700-watt monster GPUs. Using GPT, generating 100 queries for 100,000 users would cause 26 giga Watthours of electricity to be consumed. It would use 93,000 times more energy than TSF’s generator does.
- Google doesn’t care: Titles rank, but descriptions don’t. You should’ve already written a title for the article anyway, which TSF happily hands out to Google, Bing, and your friends. Search engines often toss away the description, and one’s put in place to match the search engine user’s query.
- Doesn’t fix writer’s block: If you don’t know the title or the description of an article, then how could you have even written it?
Overall, even though there could be something to gain from such a feature (or, in the case of SEO, a cool gimmick), the downsides greatly outweigh the upsides. Anyone saying that GPT is the future of the web is either unaware of the implications or are true visionaries I will never understand.
