I’ve often wondered why we haven’t gone back to the moon. After a little research into the topic, I immediately saw similarities with automated intelligence projects.
But first, the moon.
It was July of 1969, when one of the most historic events in human history took place — humans walked on the moon.
Even compared to today’s technological achievements (visits to Mars included), that was one of mankind’s most remarkable achievements. And at the time we didn’t even know if the lander wouldn’t just sink into the surface of the moon!
When the idea to go to the moon was born, none of the technology to do it was available. It all had to be invented!
And we got to the moon with a fraction of the raw processing power available in a smartphone. In fact, the entirety of NASA compute power at that time was still less than what we’ve got in the palm of our hand today.
However, what NASA did with the computers and software they invented was unquestionably powerful. NASA sent a total of 6 2-man crews to the surface of the moon. And of all the attempts, only one failed (Apollo 13). And even then, there was no loss of human life.
Why We Haven't Travelled Back to the Moon
These missions all happened over 50 years ago, and no nation has been back to the moon. There are many reasons for this, including financing, and lack of public interest.
But I believe the biggest reason is repeatability.
When you build something that accomplishes a huge goal — no matter what that is — if it can’t reliably be done again, you only get diminishing returns. The only things we can re-use from our visits to the moon are the theory and physics behind the trips.
You see, the Apollo program was never intended for long-term use. Much of the technology needed to reach the moon was destroyed on every trip (something like 2 jumbo jets worth of it!). And I'm not bashing NASA, they created the whole idea of going digital, after all...
How This Compares to Automated Intelligence
Intelligent automation is much the same. If you build a solution using deep learning or neural nets, you’ll never be able to transfer that work to another use-case. For practical reasons, it just isn't possible (more on that below):
BIG POINT: There’s a concept in the world of artificial intelligence called “The proof of concept to production gap.”
What this means is that your proof of concept — no matter how successful — will have a gap that must be traversed to get into production.
Watch a few minutes of this video at around the 8-minute mark to see for yourself:
https://www.youtube.com/watch?v=Gbnep6RJinQ
The proof of concept to production gap in every solution using deep learning / neural nets is virtually impassable. That means a lot of effort went into producing something you can’t repeat. And that’s going to kill your mission back to the moon!
So...will we ever put a manned craft on the moon? We definitely will when we've created a safe, reliable, and repeatable way to make it happen.
How Should Automated Intelligence Tools Be Built to Ensure Success?
Automated intelligence must be built using repeatable technology and approaches that transfer across departments, use-cases, and even industries.
Automation solutions that are designed to process unstructured and complex data built on development platforms like Grooper are robust, scalable, and extremely powerful.
As a result, the "proof of concept to production gap" is narrow, making repeatability a cornerstone for success.
Solutions like Grooper won’t take you to the moon, but your project will definitely be a moonshot!
