Interesting commentary from Ben Thompson about the business and profitability aspect of building digital assistants. This likely applies to many other machine learning projects as well.
Google, meanwhile, has always been a completely different kind of company a horizontal one. Nearly all of Googles costs are fixed R&D and data centers which means profitability goes hand-in-hand with marketshare, which by extension means advertising is the perfect business model. The more people using Google the more that those fixed costs can be spread out, and the more attractive Google is to advertisers.
…Secondly, though, Google has a business-model problem: the Im Feeling Lucky Button guaranteed that the search in question would not make Google any money. After all, if a user doesnt have to choose from search results, said user also doesnt have the opportunity to click an ad, thus choosing the winner of the competition Google created between its advertisers for user attention. Google Assistant has the exact same problem: where do the ads go?