Bakz T. Future
14 additional cool applications just built on top of OpenAI’s GPT-3 (generative predictive transformer) API (currently in private beta).
Links:
https://twitter.com/siddkaramcheti/status/1286168606896603136
https://twitter.com/minimaxir/status/1286100406305611778
https://twitter.com/mckaywrigley/status/1285827683776004096
https://twitter.com/plotlygraphs/status/1286079929982095360
https://twitter.com/michaeltefula/status/1285505897108832257
https://twitter.com/itsyashdani/status/1285695850300219392
https://www.youtube.com/watch?v=7Y5KsN6ehvk
https://twitter.com/amasad/status/1285789362647478272
https://twitter.com/IntuitMachine/status/1287050253103968257
https://twitter.com/mattshumer_/status/1287125015528341506
https://twitter.com/bemmu/status/1285284131656445952
https://twitter.com/ChinyaSuhail/status/1287110006370836480
https://twitter.com/JanelleCShane/status/1286327033136586754
https://twitter.com/gulan_28/status/1286251219783041026
β
βΊ Remember to Like, Comment, and Subscribe!
β
Connect with me:
Twitter – http://www.twitter.com/bakztfuture
Instagram – http://www.instagram.com/bakztfuture
Github – http://www.github.com/bakztfuture
Feel free to send me an email or just say hello:
bakztfuture@gmail.com
Small nitpick: the Legalese examples you showed were actually the training examples (that is, the priming text), not the output from GPT-3. The output was in the next photo to the right of the one you clicked π
I'll be honest: I don't understand those House List memes, and I'm a human. So I can empathize with GPT-3 trying to generate those.
please… someone make gpt-3 tell jokes
The start of natural programming languages.
So high-level, they won't be called high-level. Or programming languages.
0:31 Siddharth Karamcheti β verbs for objects
1:50 Max Woolf β don't ask gpt-3 for dating advice
2:54 Mckay Wrigley β generate an exam
3:45 plotly graphs β bar chart generation
4:24 Michael Tefula β Legalese to English
5:37 yash dani β generate financial statements for accounting from natural language via Python
6:45 Paul Bellow β RPG character backstory generator
7:30 Amjad Masad β answer questions about a program from source code
8:24 Carlos E. Perez β generate clickbait titles for Google Docs
9:17 Matt Shumer β generate code for Keras machine learning models from natural language text
11:00 Bemmu Sapponen β generate a presentation
11:48 Suhail CS β AWS DevOps commands from natural language
12:46 Janelle Shane β generate memes http://aiweirdness.com
13:18 Gulan TwentyEight β http://github.com/shreyashankar/
NumPee and not numPie pronunciation hurts my soul. Otherwise awesome video! π€ͺ
everytime you pronounced numpy i died a little
Plot twist: This entire video was generated by GPT-3. Bakz T. Future just gave it the command, "Make a youtube video about GPT-3 using my voice"
This is awesome. Not even web dev jobs are not safe from AI lol
Canβt wait to start writing beautiful stories about my Nigerian prince brother who was kidnapped by my dad the king of Nigeria, and how I need $50k to retrieve $200 million stashed in our palace.
Dating advice Nr 3: talk about the hottest trends in data science. – roflmao
GPT-5 be like: ok google, start a space mining company and make me rich
We'll finally be able to know what we're actually agreeing to on all the Terms of service!
I need the GPT-3 Virtual Personal Assistant that can access ALL (or any) of these Apps on the fly.
Ok, in the video #1 I was not (yet) compelled to join the "oh my god, we will be replaced" band. Now I am.
4:56 Excuse me, but did you not read the headline?? Those are training examples not examples of results. This was the training data, not something gpt-3 actually translated.
can GPT-3 solve halting problem?
i wanna use the api ;-;
Imagine if they baked this into a Scribblenauts game…
Ben Franklin is on the 100$ bill not Washington though, right?
Just optimize Governments already !
Is there somewhere I can use these?
I feel like Im learning to be a web developer just to be replaced by GPT-3 LMAO
I can't wait until CEO 1.0 software is released. Automation hits the c-suit.
The only really (potentially) useful one I have seen so far is the interpretation of legalese. How they got GPT-3 to do that I do not know. But can they be sure the results will always be accurate? Some legalese might not be expressible in simplified form; what does it do in that case? The reason "legalese" is used is that vernacular English is not precise.
I do not know whether creating an AI app could be done using GPT-3 as a sort of compiler.
Yes, it might help in preparing presentations, but the quality of the result would have to be determined.
These are fascinating, fun uses, but they are not practical. I am not trying to be negative or troll this discussion. GPT-3 is fascinating.
As fast as things are being iterated on right now, where will we be at like the end of the year…
GPT-3: Who is pictured on the $100 bill? George Washington.
Everyone: Very impressive!
Me: π³
Could you up the resolution or something so we can actually see the twitter videos?
That sublime text notification though