If you happen to give a bored scholar in math class a graphing calculator, they may discover a solution to entertain themselves. Most frequently, they discover a solution to hack a recreation onto their calculator, like this port of Sonic the Hedgehog 2. However not all the hacks are purely for leisure. Some focus extra on productiveness, just like the spell checker just lately developed by YouTuber ExploratoryStudios for the TI-84 Plus Silver Version calculator.
To make this spell checker work, ExploratoryStudios turned to an unlikely expertise — synthetic intelligence (AI). Not that AI is an uncommon solution to make a spell checker, in fact. Most trendy spell checkers now incorporate AI in a technique or one other. However utilizing AI may not be anticipated on this case as a result of this specific calculator is powered by a Z80 processor operating at 15 MHz, and it has solely 24 KB of RAM. Many private computer systems from the Nineteen Eighties had been extra highly effective than that.
Neuron activation patterns for quite a lot of inputs (📷: ExploratoryStudios)
ExploratoryStudios didn’t use any methods to wirelessly connect with a distant service to make this doable (as with this TI-84 that connects to ChatGPT). Reasonably, the answer, named HERMES OPTIMUS, is a customized neural community that was coded in TI-BASIC. That code runs fully on the calculator’s personal {hardware}.
That’s fairly spectacular, however when you get into the small print, you’ll naturally discover that a variety of corners needed to be lower to make that doable. At the start, the spell checker can solely function on phrases which are precisely 4 characters in size. Moreover, it might probably solely discover the almost certainly candidate for a misspelled phrase from a set of twelve choices. As such, that is actually extra of a toy than a helpful software, however it is going to assist to cross the hours in math class all the identical.
By constraining the issue on this means, a small however efficient neural community could possibly be designed. It accommodates simply 4 enter neurons, 60 hidden neurons, and 12 output neurons (one for every doable goal phrase). However even this took some work to encode every little thing sufficiently to make it match throughout the TI-84’s meager reminiscence capability. These optimizations had been adequate for inferences, however coaching nonetheless needed to be carried out on a separate pc — ExploratoryStudios calculated that it could have taken years to coach the mannequin on the calculator (if reminiscence overflow points could possibly be solved, that’s).
HERMES OPTIMUS could also be little greater than a toy, however by coding the neural community manually in BASIC, it’s a good solution to be taught extra about machine studying, so it’s well-worth the train if in case you have an previous calculator sitting round. In case you are eager about studying extra about how AI can run on some significantly outdated {hardware}, you must try my TensorFlow Lite for Commodore 64 and Commodore 64 AI Picture Generator tasks.
