This article is more than 1 year old
Free gift for all readers: Google's AutoML launch translated into plain English (where possible)
It's an image-recognition thing
Google today tore the covers off something called Cloud AutoML, a new service that's part of its "mission to democratize AI."
The technology is touted as artificial intelligence that can design artificial intelligence. In other words, rather than build your own machine-learning software, Google's Cloud AutoML will take your training data, figure out what you want, and automatically generate intelligent code for you.
In reality, it's an image-classification system.
"Only a handful of businesses in the world have access to the talent and budgets needed to fully appreciate the advancements of ML and AI," cooed Fei-Fei Li and Jia Li, chief scientist and head of R&D respectively at Google's Cloud AI, in a blog post.
“There’s a very limited number of people that can create advanced machine learning models. And if you’re one of the companies that has access to ML/AI engineers, you still have to manage the time-intensive and complicated process of building your own custom ML model.”
To translate: you don't have any decent machine-learning programmers. Don't even think about doing that stuff yourself. Google has them all. Google knows best. Just let Google do the AI coding for you. And please enter your credit card, here.
At the moment, Cloud AutoML can only deal with computer vision problems.
Businesses are invited to upload images to train a neural network that is tailored to their interests. For example, a clothes manufacturer may want to feed Google’s cloud pictures of t-shirts, sweaters, skirts, and so on, to create a system that can identify different types of clothing in stock.
The images need to be labelled so the software can learn to recognize objects. Google offers a “human labeling service,” or users can hand label the training data themselves.
AutoML started as an internal project within the web ad giant's research wing dubbed Google Brain. It boffins were exploring ways to automate the process of designing neural network architectures using machine learning. They used a mixture of evolutionary algorithms and reinforcement learning to come up with what's called neural architecture search, which is basically software automating the task of crafting and testing new machine-learning models.
Boffins foresee most software written by machines in 2040READ MORE
Now, it appears neural architecture search and transfer learning are being used together to power Cloud AutoML, spitting out customized models on demand. It’s interesting to see exciting research deployed in production, but it’s unclear how much of it is genuine intelligence – and how much is templated code and models that are produced by humans and selected by bots for customers.
The above blog post talks about using drop-down menus to build a bespoke AI, which is a bit of a giveaway. Calling it Cloud DropDownML probably wasn't dramatic enough.
Does Google create from scratch a new model for every customer? Or does it slightly tweak a previously trained one for different use cases? It’s also unclear how much this service costs. Is it affordable for smaller businesses? Are customers billed by the complexity of the model needed? A spokesperson for Google has yet to get back to us with any answers.
Maybe if you have a large enough credit limit, you can find out the answers yourself direct from the internet advertising network. ®