GPTZero Upgrades Its Plans for Schools
Ed Tian launches "GPTZeroX", it's like a B2B grade plan where GPTZeroX may be leased to institutions.
Hey Everyone,
I have covered Edward Tian’s rise as a student builder founding GPTZero for educators, now he’s doubled down on how this might look.
GPTZero, as you may know, is a tool that can sniff out AI-written text, has been accessed by 80,000 people since its January 3 launch, said Edward Tian, who wrote the app in a Toronto cafe.
Tian is part of a new generation of A.I leaders with both Chinese and American connections. The grandson of a Tsinghua University-trained electrical engineer, Tian is months from completing his double major in computer science and journalism.
In a recent post on his Substack, he clarifies and gives new insights on his GPTZero tool’s outlook:
Core Updates:
two exciting new features and a new model, GPTZeroX, are rolling out next week. See tech updates for the deets :)
we’re in talks with school boards and scholarships to service GPTZeroX to institutions, including potentially 300,000K+ applications for one of the world’s largest student scholarship funds. If your organization might be interested, please let us know
Tech Updates
New feature: Word and PDF uploading implemented. Demo in heading. [Expected roll out this Monday]
New model: GPTZeroX responds to the feedback we’ve received, where the beta performs poorly with edge cases, shortened text, and AI generated writing that’s been lightly or adversarially edited afterwards. The improved model successfully services these cases, and adeptly handles text that combines both AI and human writing together.
New feature: highlighting areas of a text that are AI involved. If a text is completely human, we’ll tell you right away. If GPTZeroX is not sure, we’ll highlight the portions of a text that are likely AI generated instead, and transparently, provide you all the probabilities. [Expected roll out Thursday]
He runs a Facebook Group for Educators and teachers you can access here. It’s up to about 250 users.
Tian’s tool is a bunch of a new wave of tools that tries to look at ChatGPT’s use in an increasing world of plagiarism by A.I. Spam and copying styles is just the beginning of what these writing tools will do to the internet, many are making dire predictions of what content might look like even by 2025 online.
ChatGPT can be a useful tool for generating text and responding to prompts, it can also be a problem for teachers and editors when it comes to plagiarism.
Edward Tian has been getting good coverage for his A.I. tool GPTZero, including coverage by news outlets, including the BBC, SCMS, Information, WaPo, and WSJ. Not all are as lucky. Pro tip, there are many aggregators for A.I. Tools including:
And honestly dozens if not hundreds of others.
Newsletters are also getting jacked by mentioning A.I. tools, a quick-fire way to go more viral.
Edward Tian
While there are so many builders and solo-entrepreneurs let’s cover more about Ed Tian.
Featuring the tagline “humans deserve to know the truth”, the app is aimed at educators, and can assess whether a corpus of text was written by humans or artificial intelligence (AI).
GPTZero highlights the myraid tools and issues that ChatGPT is creating online even in just a limited demo.
AIgiarism
It is increasingly being used by students as the crutch for writing their assignments and essays, a practice dubbed “AIgiarism” that has teachers in many US schools up in arms.
The SCMP got some good quotes:
ChatGPT is the Pandora’s Box of social media technology, as it comes with pros and cons like any new technology, said Tian. “While it’s a brilliant innovation, there are also lots of downsides too.”
Edward Tian got an internship at Microsoft at just the right time.
How does the app work?
GPTZero uses two variables to identify if text shown to it is written by a bot or human: perplexity and burstiness. Perplexity is a measurement of the randomness in text. If a corpus of text is unfamiliar (perplexing) to the bot, then it is more likely to have been written by a human.
As a builder he is in talks with school boards and scholarships to service GPTZeroX to institutions, including potentially 300,000K+ applications for one of the world’s largest student scholarship funds.
With the New York school board banning ChatGPT, it’s a peculiar situation with ChatGPT being used for creative and nefarious uses. Others are arguing we should be adopting it into the curriculum itself.
ChatGPT is supposed to go live in March on Bing, Microsoft’s $12 Billion Search Ads business, and competitor to Google. Meanwhile companies like Amazon, Microsoft, Apple, ByteDance, Netflix and others are getting more into Ads.
Microsoft may be able to adopt ChatGPT and its features into its software products and make its moat even stronger, it’s an antitrust moment in its planned bid to give OpenAI $10 Billion more, for a 49% stake in its business.
The future is not clear at all with a wave of Generative A.I. Even Stability.AI (hosted on AWS) is looking into a ChatGPT competitor.
The OpenAI splinter A.I. labs are now into their next generation from Adept AI. This suggests A.I. labs will keep spinning out, and the push-pull of BigTech buying up first-movers and the movement of really democratize A.I. is the new bifurcation of A.I. R&D. Can lone builders with tools build into legit startups and can A.I. labs make progress without the involvement of big sponsors, it all remains to be seen.