Giving an Old Typewriter a Mind of Its Own with GPT-3

Published by Reblogs - Credits in Posts,



Skip to content

Giving An Old Typewriter A Mind Of Its Own With GPT-3

There was an all-too-brief period in history where typewriters went from clunky, purely mechanical beasts to streamlined, portable electromechanical devices. But when the 80s came around and the PC revolution started, the typewriting was on the wall for these machines, and by the 90s everyone had a PC, a printer, and Microsoft Word. And thus the little daisy-wheel typewriters began to populate thrift shops all over the world.

That’s fine with us, because it gave [Arvind Sanjeev] a chance to build "Ghostwriter", an AI-powered automatic typewriter. The donor machine was a clapped-out Brother electronic typewriter, which needed a bit of TLC to bring it back to working condition. From there, [Arvind] worked out the keyboard matrix and programmed an Arduino to drive the typewriter, both read and write. A Raspberry Pi running the OpenAI Python API for GPT-3 talks to the Arduino over serial, which basically means you can enter a GPT writing prompt with the keyboard and have the machine spit out a dead-tree version of the results.

To mix things up a bit, [Arvind] added a pair of pots to control the creativity and length of the response, plus an OLED screen which seems only to provide some cute animations, which we don’t hate. We also don’t hate the new paint job the typewriter got, but the jury is still out on the "poetry" that it typed up. Eye of the beholder, we suppose.

Whatever you think of GPT’s capabilities, this is still a neat build and a nice reuse of otherwise dead-end electronics. Need a bit more help building natural language AI into your next project? Our own [Donald Papp] will get you up to speed on that.

[via It’s Nice That]

26 thoughts on "Giving An Old Typewriter A Mind Of Its Own With GPT-3"

    1. Made me think of the "word accordion" from Stephen King’s Tommyknockers.

  2. Instead of an AI people should learn again speaking to other people normally on the internet. An AI is no more than statistics + what other humans wrote on the internet. There is no intelligence, it just utters what is statistically the next chain of words in that context. I cannot wait to go to SawCon 2023 and meet real people.

    1. The Commenter Formerly Known As Ren says:

      SawCon?
      Is that a convention of fans of the SAW movie?

      1. Something like that :3

      2. They also play games all the time ;)

      3. Sawcon deez nuts

    2. Stop harassing Al Williams, does the best it can…

      1. "Al" Williams or "AI" Williams. Coincidence? Hmmm.

    3. Have you tried talking to most people on the internet? They’re terrible!

      Anyway, saying "blah blah learn to talk to other humans" sounds an awful lot like "go play outside, video games will rot your brain" or "TV will rot your brain", "reading will rot your brain", ad nauseum.

      Trained "AI" models will have their uses and before long they’ll be normalised like everything else we do.

      1. Oh my idea was more of the lines of: Being excellent requires effort.

        Talking normally to people encourages most of the to talk normally to you. The majority of people is sane and not a troll who gets his kicks out of causing other people misery.

        An AI will always be unhinged or a pretender, since it has no real human motivation. All you get is the statistically most probable answer for the context of your subject. Either way people on Hackaday are luckily more laid back than in other online communities.

        Real people can be random, which is much more fun! If an AI fakes randomness it would probably sound schizophrenic.

  3. I need to see the Selectric 251

    1. A little hacking history, the IBM Selectric series was the typewriter with the type "ball" element. It was completely mechanical and had 7 bails on the bottom that did all the encoding from key to print. The available printers of the day were low quality, so the trick was to put a bottom on the typewriter with 7 solenoids to pull on the bails. You get the first hacked letter quality printer. Even better, back before PROMs were practical, National Semiconductor used to have a masked ROM that translated ASCII to Selectric. It would output the proper bails to print the ascii character.
      The prop that was the Selectric 251 was probably one of those modded units.
      I had to do some EMI work on some for the Gov’t once, so it was well know hack.

      1. Or just one of the many Selectric-based terminals available unmodified.

      2. I was there, I remember when IBM Selectics were used as printers. Nothing said magic like printer enclosures. I think we used WordStar at the time and of course WordPerfect. It was like a warzone when these things would crank out the pages.

  4. Id have rather seen, speech to text input with gpt3 to mechanical output. Fake sentient type writer style.

  5. Gregg Eshelman says:

    How about that typewriter for children that had a switch to encode the text? It remapped the characters. https://www.cryptomuseum.com/crypto/mehano/barbie/

  6. Oh my idea was more of the lines of: Being excellent requires effort.

    Talking normally to people encourages most of the to talk normally to you. The majority of people is sane and not a troll who gets his kicks out of causing other people misery.

    An AI will always be unhinged or a pretender, since it has no real human motivation. All you get is the statistically most probable answer for the context of your subject. Either way people on Hackaday are luckily more laid back than in other online communities.

    Real people can be random, which is much more fun! If an AI fakes randomness it would probably sound schizophrenic.

    1. Whoops this was meant for Erik T. Replying in Firefox is broken after so many years .w.

      1. Maybe an AI chat moderator will fix the chat for us :)

  7. I did something similar with ChatGPT very recently:
    https://vm.tiktok.com/ZMY1KNRcU/

    The typewriter I used was a brother TC-600, an obscure portable thermal typewriter that already comes with a serial terminal interface.
    I made a python script that basically wraps https://github.com/labteral/chatgpt-python into a serial console.

    It’s really impressive to type in a prompt and seeing the response being printed like magic!

  8. >ChatGPT enters the building

    It is possible to give an old typewriter a "mind of its own" using GPT-3, by integrating the GPT-3 API with the typewriter’s mechanical systems. This would allow the typewriter to respond to prompts and generate text based on the input provided to the GPT-3 model.

    To do this, you would need to have a basic understanding of mechanical engineering and programming, as well as access to the GPT-3 API. The process would involve connecting the typewriter’s mechanical systems to a microcontroller or computer that can communicate with the GPT-3 API.

    Once the connection is established, you would need to write code that interfaces with the GPT-3 API and controls the typewriter’s mechanics. This code would need to process the input given to GPT-3, and translate that into commands that the typewriter can understand, such as which keys to press and when.

    It’s worth noting that this would require a lot of engineering and programming work, as well as a deep understanding of the GPT-3 API, and may not be a easy task. Also, this typewriter would be a one-of-a-kind machine and may not be fully functional as a regular typewriter.

    >ChatGPT exits the building

  9. This looks like a job for Eliza!

    AI but period appropriate too!

  10. This somehow reminds me of "Naked Lunch"…..

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Search

Search for:

Subscribe
Copyright © 2023 | Hackaday, Hack A Day, and the Skull and Wrenches Logo are Trademarks of Hackaday.com | Privacy Policy | Terms of Service
Powered by WordPress VIP