Smart Text

AI ChatBot No One Likes ;-)... No Privacy (A Problem We Ran Into And Made A Feature)

Comment 2

Inspiration

We wanted to create something that was truly useless. Something that made you go, "Why is this a thing?" Well, that is what led use to create a useless AI.

What it does

Our AI Bot takes all the letter in the alphabet, turns them to ASCII characters, jumbles them up and tries to link every ASCII character to its corresponding letter. In other words, our AI Just Says What You Say

How we built it

The majority of the coding was done in Visual Studio Code with Python. One of us would run the code to train the bot, then that data would get sent to another person to train the bot. That way, the training wasn't on one person, but the whole team.

Challenges we ran into

A major challenge we ran into was communication. Our most experienced programmer lives in India, so the time-zone difference was rough. We used Discord to keep up with each other as well as Google Meet. We also ran into some problems turning the characters to ASCII, as we wanted the Data in a int array, rather than an string array. 

def get_ascii(arr):
    ascii_list = []

    for char in arr:
        ascii_list.append(ord(char[0]))

    return ascii_list

Accomplishments that we're proud of

We managed to train the AI to 90-95% accuracy, which is quite impressive, considering it roughly went up by 1% every 10 minutes. For the majority of the team, this is their hackathon, so actually producing something is remarkable.

What we learned

It's rather what we haven't learned... Some of us had never done anything related to machine learning before doing this project, but we've come out so much smarter than we were. We had a session where we just talked about how it works, from neutrons to weight to hidden layers. I, Oliver, had never done any kind of python programming before, so being thrown in the deep-end was a great opportunity to learn and explore.

What's next for Smart Text

Smart Text is meant to be an impractical joke. Using AI to decode some ASCII characters is just not efficient, nor does it really have any place in the future. It was a fun project, but it's not realistic. However, that is not to say that we won't use the base components in future Hackathon. As a matter of fact, a couple of us are attending a Quantum Machine Learning Hackathon in a weeks time, so re-using some of the code isn't out of the question.

Built With

Share this project:

Updates

posted an update —

The Time Has Come... Submission

Update 5

It's time... submission time. I would like to take this opportunity to thank everyone in the team for their amazing effort in this. Sushant, you carried the team and it wouldn't have been possible without you! Best of luck to everybody, and have fun checking out our goofy hack!

Log in or sign up for Devpost to join the conversation.

posted an update —

It's Domaiiinnnnnn Time

Update 4

We've just registered our domain with domain.com. We will be developing a website to go along with the AI, so that everyone will be able to see the quirkinesses of our project.

SmartText.tech It's not currently online, as we're still working on it!

Log in or sign up for Devpost to join the conversation.

posted an update —

Decision time!

Update 2

So we've decided to do option #3, machine learning for decoding using ASCII. If curious, here is a snippet converting characters to ASCII:

def get_ascii(arr):
    ascii_list = []

    for char in arr:
        ascii_list.append(ord(char[0])) # char[0] so it doesnt grab "\n"

    print(ascii_list)
    return ascii_list

Log in or sign up for Devpost to join the conversation.