As I mentioned in a previous post, artificial intelligence (AI) is one of the hottest topics going around the traps. Developers are always keen to jump onto the latest trends and topics, so naturally assume that they can bend their minds to the latest hype. Amazon, Google and Microsoft down play AI by offering pre-packaged solutions that can be bolted onto new or existing applications. I say down-play because they all but eliminate the need to focus too much attention on the analysis of data, building and validating an appropriate algorithm/model. For example you can build a custom vision service by throwing up a bunch of images, appropriately tagged, to train a model. Thus a developer with minimal AI skills can build and integrate such a service.
For developers who really want to build AI solutions, they can do a basic AI course (for example Microsoft’s Professional Program has an Artificial Intelligence track) but does this mean they’re an AI expert? What if they go and learn Python or R, does that make them an AI guru? The short answer is no – they’re still a software developer, not a data scientist.
If you’re a company that wants to add some AI to your application you should look beyond the basic claims by development companies that they “do AI” and verify what they mean. There are specialist data centric companies out there, like Analytics 8, that have a history in working with big data – if you have vast data repositories, this is where you should start. Once you’ve defined what your AI solution is, then you can engage with a developer to integrate the model into your application, website or bot.