What are the dark sides of a career in AI/Machine Learning?

  1. Makes you more sarcastic
  2. Overworking and looking for answers to life, universe and everything (42 is just biasing aka y=42)
  3. It makes one write long answers totally aimed at making you understand why you chose the random.seed(42)

Lets turn the pages in our good fellow computer science’s history and what all things researchers and developers in AI have gone through.

This is what the people who discredited perceptron algorithm must have said which was invented in 1957 as it wasn’t the answer to AI for it had given hope to the world that we will soon have robot slaves. There were no developers and AI was strictly a research topic which seemed super futuristic.

This was around 12–14 years back when my brother was doing his CS undergrad when AI by Rich and Knight was considered a bible for it broke down how machines can be made to take decisions by converting the human language and problem statements into a simple logical statement and a knowledge base.

(By this time, neural networks had grown by leaps and bounds for CNN, RNN and Boltzmann Machines were very popular in research circles but they never got out into the industry)

Research community flourished for the internet boom allowed R&D units of AT&T, IBM, Microsoft and many other top corps. got best minds to tackle real-life problems giving rise to MNIST database and other algorithms to be discovered.

Developers still looked at logging user interactions and behaviour and found it difficult to replicate research to production in a language know to them.

GANs was introduced.

Coding became easier when ml/dl libraries backed by Google and other big research houses written in an even easier language (Python).

Computing became faster courtesy GPUs and Cuda programming.

Also, data analytics and data science became new, rebranded terms of glorious amalgamation of CS and Statistics.

The research was booming. Neural Networks were back with a bang and computer vision like a big brother introduced the commercial world to deep learning applications like self-driving cars.

Developers still believed they had time to catch up to the research as it is still not production level and scalable.

Present Day -

People have gone bonkers behind the terms AI, ML, Deep Learning, NLP.

Hiring has been an all-time high.

Data scientist, the sexiest job of the 21st century.

People in testing want to do automation and learn ML/DL to increase their repertoire and shift gears.

Developers and code maintenance engineers treat AI as their ticket to freedom from a mundane job courtesy tons of MOOCs and tutorials and higher level APIs like Keras and putting simple implementation on medium articles and LinkedIn and treating it like they have cracked the human genome.

Future :

AI that makes new AI, maintains itself like a human body. Totally fed up of imposters posing as a human for they are just the sheep following the herd. The people who control AI are the new alpha and omega of the world and can destroy cities with just a snap.

Researchers are still mulling over less funding and not getting the new quantum supercomputer to play flappy bird.

Developers and testers still have

Looking for opportunities in ML/AI

in their Linkedin bio and children are taught if else loops before they even learn to speak…of course, they are Sophia’s children.

TL: DR;

Don’t run behind the herd. If you are genuinely interested in any of the subfield of AI, now is the right time to work your ass off and get funding or your product out which will get you through your whole life. Else, every field gets you paid and successful if you know what you are doing.

Testers and Developers, you guys don’t need to curse your jobs! If you didn’t like working as a developer/tester then it will be a hell lot frustrating to work as ML engineer or AI for you will be paid a bit more than what you are making right now and will be expected to develop a mobile /web app, data cleaning, data modelling, dev-op-ing your way with AWS and other cloud services to finally test your model at least once which still needs to be scaled and optimized to run on a Raspberry Pi. And then you will realise what people discovered when they found out that perceptron was only good at solving linearly separable problems and nothing else. And we all know real life problems are never 2D. if you wish to know about Artifical Intelligence you can visit Learn Artificial Intelligence Fundamental

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store