Salesforce’s artificial intelligence division has been busy over the last few months. Since the release of Salesforce Einstein in September they’ve published four research papers and claimed to have remade the way that computers read and understand language in the process.
That’s a hefty claim but, as usual, Salesforce has the data to back it up. When they introduced Einstein in September, Salesforce promised to embed AI across its cloud services.
Salesforce submitted several of its papers to ICLR 2017, a machine-learning conference in April next year. Submissions won’t be reviewed and approved (or rejected) until February 2017. Until then, all we can do is trust the CRM giant and test things out for ourselves.
What does Salesforce Einstein do?
Einstein isn’t so much of a product as a set of intelligence functionality that underlies the entire Salesforce platform. The idea behind Einstein is to build a foundation on which the company can continue to add new capabilities in the future.
AI moves at a rapid pace and the science gets complex quickly. But the gist of Salesforce’s claims and the current capabilities that Einstein has to offer, such as natural language processing (NLP) tools, is that they’ve found more ways to compare documents to each other and questions users ask so that the AI can better understand what people are looking for and help them find it.
Ultimately, Einstein aims to reduce the amount of time people spend entering information because it’s going to do most of the work for them, and because it’s intelligent, it will provide you with meaningful suggestions about what to do next. Think of it like a guide that’s willing to make a lot of the hard decisions for you.
“AI can surface the most important information to the top and say, ‘hey, this is what you need to do next,’ so the salesperson can do what they do best, which is connecting with the customer, and not spend so much time on entering data,” Shubha Nabar, director of data science at Salesforce explains.
Salesforce Einstein looks to the future
The biggest promise Salesforce users may have to look forward to is a mouthful of a method called Dynamic Coattention Network for Question Answering or DCN for short. DCN focuses on the interplay between documents people are searching for and the questions they are most likely to ask.
Salesforce’s system learns about documents by scanning them and taking hints about what’s in the documents by the questions people ask about them. The AI currently holds the No. 3 spot on the SQuAD leaderboard—SQuAD, the Stanford Question Answering Dataset is a newly developed reading comprehension test for computers based on articles from Wikipedia that launched in June.
Another noteworthy claim of Salesforce Einstein is called the Pointer Sentinel Mixture Model which basically uses context clues when reading words it doesn’t know. The system can “point” to unknown words and figure out what they are based on how they are discussed in a document.
Einstein has the strength of 10 men
Perhaps most interestingly of all is Salesforce’s claim that they have developed a single AI system that can do the work of several language analysis systems. Basically, Salesforce is saying that their new system has the strength of 10 men in that it can handle a search bar, live chat messenger, FAQ and translations system all at once.
We’re still in the early days of artificial intelligence, but we should start to see the technology mature over the next decade, and for Salesforce users, that could mean working faster, smarter, and knowing the best move to make before anyone else does.
Einstein is still young so only time (and Salesforce users) will be able to tell how much the AI is truly capable of.