We still don’t understand how profound AI capabilities are
Google’s CEO hit the nail right on the head when he said that AI capabilities are more profound than fire or electricity.
We can do amazing things with today’s wide array of Machine Learning and Deep learning algorithms. As per my perspective, we are constrained only by our imagination. But one additional “imaginary” constraint that I often hear is – But the supporting tech infrastructure is not there yet. In my mind, that is not something that should stop you from formulating initial solutions and running pilots. Why ? Because the process of identifying opportunity areas, developing algorithms, running pilots etc. in itself is time consuming so by the time you are done with them, the technology will be in place.
Technology infrastructure will not be a constraint
Unless your vision is extremely futuristic, do not worry about technology infrastructure being a challenge. Technology is progressing at such a rapid (sometimes frightening) pace that by the time you have a tested algorithm, you will have the technology infrastructure in place. So at a very high level, here are few steps that you need to start taking, without worrying about things like technology maturity to support or real time data availability:
- Map the end to end current state processes
- Define future state, considering your future state vision and operating strategy
- Identify types of algorithms/solutions you need to build for the future state (assuming that the supporting technology platform and hardware will not be a constraint)
- Classify algorithms into buckets like custom build, off the shelf etc.
- Run pilots for all the areas where the algorithms fall under “custom build”.
I can assure you that by the time you are done running the pilots, you will have the tech infrastructure to make your future state vision possible. For pilots you can use historical data and placeholder entities in place of tech hardware but DO NOT wait for tech infrastructure to mature in next few years. Things in the world of AI are moving so fast that if you wait, you will never be able to get back into the race.
A real life example of Technology catching up with algorithm
In August of 2019, I developed few algorithms that would essentially connect with Warehouse systems and do Smart Yard management planning. The vision I had required tech infrastructure, in terms of a Digital YMS system that was not available at that point. I anyways tested my algorithms with assumptions and applied for a patent.
Few months after applying for a provisional patent, vendors started coming into the market offering exactly the Digital YMS solutions offering the tech platform that I envisioned my algorithms would use (See screengrab of one such product below).