How to corrupt an intern robot

Although titled: "How to corrupt my intern robot", there is a LOT more information contained within this video than you would actually expect.  I would highly recommend a full watch, or zoom FWD to 11:39.

Okay, so Misty the robot is pretty cool  CLICK HERE TO SEE / READ / VIEW MORE

I think its great that they've made different ways to code the robot, from 100% beginner using BLOCKLY (a person fav. of mine for non-coders to get to understand coding)

You also have the option to code using the Python SDK, Web APIs and C# SDK; which is pretty cool.  So, you can go from total novice through to expert coder, well, as expert as calling HTTP APIs or using Visual Studio is...no C-code level for the hardcore.... except, oh wow!

https://docs.mistyrobotics.com/misty-ii/robot/misty-backpack-for-arduino/

They have a backpack module that is an Arduino, so you can extend and enhance the robot yourself...and use C (coz that's what Arduino is coded in), so it looks like they've covered all bases.

However, at about £2500 (currently), it's a little bit beyond the price point for a novel toy-esque plastic robot.  I get that it has funky sensors, servo's, wheels, screen and PSU all built in etc... but the price is a little steep for "having a play around", especially considering you'll be coding all the logic in an IDE on a laptop mostly - which if you're doing that, you can run/execute all the code pieces without having to use the actual robot as the UI output - facial expressions and moving 2 servos to move arms up/down isn't exactly floating my boat for that kind of money.  Don't get me wrong, I love it and I would like to have one on my desk - however, I can achieve the same sort of thing using a Raspberry Pi with my Google HAT and some Python code, of which I have already done and used as an offline NLP language translator:

I digress though.  I think Misty is cool.  Also reminds me of the HANSON ROBOTICS LITTLE SOPHIA robot that I KickStarter backed (what seems like eon's ago!) - maybe it will get a commercial release next year?

So, the reason why I liked the VIDEO above, was partly because of Misty the robot, partly because it started to explained GTP-3 and GTP-J6B (alternative "free" option) and MOSTLY because of this section of the VIDEO (11:39 --> )

I absolutely 100% agree with the statement being made here - and I work with, so called, "experts" in this area and some do actually think this stuff is way more advanced than it really is (I love the comment about, "no matter when you watch this video, the technology you're thinking of will always be 10years away" - so true)

This section on NLTK is a WONDERFULLY quick & simple way to explain how NLP works - yes, it is overly simplified, but once you understand that is actually how the basics work, you can start to understand pretty much how everything that uses NLTK works:

VIDEO (12:40 --.>)

and here's an example:
"How many mice does it take to screw in a lightbulb?"

You can see how using a few lines of code and the NLTK library, you can start to extract out the meaning (Intent) and the values (Entities) that you can then use to "understand" what is required and then go and use if...then...else... code to go find out the data to construct a response.  See....no magic..no real AI either.

In a nut-shell, if you've come from a chatbot (yes, I point blank REFUSE to call them Virtual Assistants - they are not and therefore I shall not call them that) background, you can see that it's not that difficult a wall to climb to get to understand Natural Language Processing (NLP) using toolkits that are already available - if you're a bit sadistic like me and end up writing your own NLTK then you've either got too much time on your hands or you're attempting to make a product that you can sell to other people - but, it has to be better than the "free" open-source libraries, of which it won't be... so you'll end up just putting a wrapper around the "free" open-source libraries and making your wrapper have a unique selling point and then call it "Alexa", ah no, that's already taken, you'll call it "Hey, Bob"...

I was once forced to make a 12min work video, entitled, "AI - there is no magic", I found it amusing and I even filmed it in my "garage", maybe one-day it'll see the light of day on the internet - it still holds true though, there is "no magic", it's just lines of code that a human has had to put together and test / debug enough to get it out the door and used.

Talking of work/work, here's some pretty cool new libraries that IBM have extracted out of their Watson products and have made available to developers, so that coders can now pretty much do what was explained above - but use the benefit of billions of dollars worth of investment and time/model-training/effort for use in your own NLP apps.  Seems daft to start from scratch when you can be "lifted onto the shoulders of giants" to accelerate your own work:

https://dataplatform.cloud.ibm.com/docs/content/wsj/analyze-data/watson-nlp.html

right...back to doing something useful now...