By now, most people have heard about AI. AI has taken the tech world by storm. However, few realize that this is decades in the making.
In the 90s, a professor had given me a book on neural networks and told me this subject is going to be huge. It has taken several years before this technology is ready for prime time with tremendous innovations in Hardware, Cloud & Software.
I heard this saying a few years ago – “With AI, Hard Problems are Easy and Easy problems are hard!”. While AI promises to solve almost any problem – it’s really important to choose the problem you want to solve carefully. For example – it has taken many years for Siri to understand speech properly – and even now it struggles with accents. This is an example of a problem that is easy for humans to solve but difficult for machines.
Detecting spam in email or translating text in many different languages real time – while not perfect – are examples where machines are better than humans.
There have been many examples and papers written where AI can do amazing things. One such example is where AI can detect cancer in early stages with much higher accuracy than the best doctors. Browsing some interesting applications source code on Github, a large part of the work can be accomplished by relatively very few lines of code. The actual hard work is abstracted away in libraries or in the data model.
I’ve always believed in AI and on my team, a few years ago we designed a course to get interested engineers quickly ramped up on AI and can build applications. I went through the course and also learned about the experience of others. In most cases, a large part of the work in AI is about cleaning data and making sure you have enough samples for AI to learn. There are of course more interesting things like vision libraries and generative AI but as an engineer the work here is quite different.
After a lot of thinking – I come back to the AI Paradox and which problem is worth solving with AI?
AI has the potential to increase our productivity 10x or more – that does mean many jobs are at risk but it also means that a lot more is possible. The difference between “good” and “great” becomes lesser because the barrier to get to “good” is much lower now.
The future is very exciting for engineers working on AI as there are endless possibilities and the recent advances and public interest is only going to bring more exciting opportunities.