Thoughts about AI market state in 2019 (after o'reilly artificial intelligence conference)

While I doubt very much the "training" value of conferences, I find it very useful to visit 1-2 conferences a year to learn more about trends in the industry. You can get a pretty good understanding of the current market state just by watching the kind of companies & people presented.

I've been watching the state of AI market for the past few years on various "AI New York <conference name here>" conferences and there is a clear evolution of AI market that I see.

The era technology(s) development

The AI industry started to grow with "hey, look at this cool trick I can do with Machine Learning!" (yeah, marketers used "Machine Learning" back in days instead of broad "AI"). ML demos looked really cool and promised to solve any problem in the world.

The era of business value recognition

If you google for "ML/AI adoption problems" in 2016-2017, then one of the first things every article would start with is "businesses are certainly interested in ML/AI, but struggle with the vision".

So, the next logical step of AI market evolution was "we are not just doing tricks, but solving real business problems". The era of business value has begun. Technologies got more mature, the industry learned about typical use cases when AI is actually useful and helps to make money. This is the time of the rapid growth of AI startups. Every single domain and even individual feature were reinvented by somebody using the power of ML/AI technology.

(hopefully) The era of AI as a commodity

This year, I went to O'Reilly AI 2019 - New York conference. From what I can see, the market shifted into building general-purpose platforms. Players that I've seen in the last few years either struggle to survive, or acquired by larger players, or expanded the original product into a broader vision platform.

I think there are a few reasons for that:

  • building an AI platform for a specific business use case is still quite expensive. Only large players have such massive data sets and userbase that 1-2% optimization worth investments into AI. There is clearly a market for general purpose platforms that businesses can quickly adapt and customize
  • in the past few years industry accumulated knowledge about typical AI applicability. This makes it possible to generalize the solution and build a new generation platform with experience and patterns of solving typical problems using general purpose tool-set

Instead of conclusion

The battle for the right be The AI Platform is ON. Likely there would be multiple winners and market consolidation. I think fairly soon general purpose platforms will provide cheap commoditized service of ML model templates and their execution. Those models will be more or less interchangeable and could be taught to anything business may need. Think of these AI platforms as cloud solutions right now. They have more or less the same pricing model. You run applications on them and that's where the real cost and money are -- on a domain and application level.

So, if you are an engineer, then learning how to code AI is must have these days. If you are business then stay focus on your vision. Affordable AI is coming.

#TheAIConf #machinelearning #AI #artificialintelligence #machinelearning

要查看或添加评论,请登录

社区洞察

其他会员也浏览了