Artificial intelligence and venture capital: critical opportunity time

It's an exciting time to be investing in tech: artificial intelligence unlocks some of the promise to reduce tedious labor and engage the spectrum of human imagination.

artificial intelligence and venture capital

I was having lunch with an engineer friend today. The conversation quickly went into the most investable companies in the region, and he asked me why I’m such a fan of artificial intelligence. It’s not just the cool factor–it’s the scale factor.

Artificial intelligence and venture capital

AI is not a special breed of software as much as it’s an approach that can be a small or a large part of any system. That approach is about creating a learning system, in contrast with building a nonlearning system (or subprocess).

So then the question gets a little simpler: do you want systems that self-heal and learn over time, or do you want assembly-line software productions that use people for incremental improvements? I prefer keeping people out of scale as a process, as much as possible. Let’s look at how software is built to see why.

Traditional way of developing software

You have a problem, and you start with the problem.

You think hard about the rules and variables that can define the problem.  You build software that wrestles with these variables so fast it looks like magic. But still, your solution isn’t perfect, which is what you expected. So you monitor and measure the gap between your solution’s performance and the problem.

As fast as your budget can afford, you write, test and deploy new code to narrow the gap between your actual solution and a theoretical perfection solution to the problem. You iterate as fast as your developers can manage. The more reality your system encounter, the faster your list of bugs and needed fixes grows.

The imagination of the engineers is a critical factor, but so is the manpower–how many developers you can throw at the problem makes a big difference.

The automated intelligence way

You have a goal outcome in mind. You frame all the parameters your team can think of that govern that goal’s achievement. You put data into your system– a lot of it, to see how far off you are from achieving the goals. But still, your solution isn’t perfect, which is what you expected. So the system monitors and measures the confidence it has between goal achievement and its actual performance. You’re pretty far off, which is what you expected. With lots of data, the system iterates as fast as your processing can manage. (No hands, mom!)

Over time, it tunes its initial equations to start achieving the goal more often. Soon, if the framework is strong enough and it has enough information to work with, it achieves the goal most of the time. Then you let it play with reality and it gets even more data. If it’s a strong framework, it keeps improving itself. If it’s too fragile, it produces garbage and you shut it down. The more reality a well designed system encounters, the better it gets at hitting its goals. The richness and diversity in the imagination of the engineers, and the richness and diversity of quality data, are critical shaping factors.

Investing capital in artificial intelligence

The AI opportunity for early stage tech investors is that it takes longer to “gestate” than the old way of developing systems, but then can scale more rapidly. You have a seed stage when you build the core system–that part is the same.

But with automated intelligence approaches, you also have a learning and tuning phase that can be quite lengthy (and really never ends). This means early stage capital is needed even more to build a learning system than for other types of systems. Following the development cycles years to out their logical conclusions, proven learning processes will gradually replace more linear variable-passing code elements so that developers can assemble smart systems, further reducing the dedicated human labor involved in coding, and further prioritizing the need for  imagination and design engineering at the founder’s stage to create the most competitive systems.

Artificial intelligence is today

Many of these learning processes, like voice and facial recognition, are being tuned today. You’re probably a daily user of voice recognition natural language processing if you text by speaking, use Alexa, or Siri. It’s an exciting time to be investing in tech: we’re on a new frontier in how the solutions that software can offer us are built for scale.