It’s well known and understood that Ray Kurzweil is working at Google to create an AI that can understand you, your needs and most importantly to understand your context and anticipate those needs. Basically it’s an artificial intelligence that will ultimately replace search and be your assistant. Say hello to Allo. Allo features an AI assistant that learns from your interactions. Allo is the first baby step to the new algorithm. If you think Penguin and Panda are sophisticated, take a look at what’s on the horizon. Allo is more than just a messenger app.
Search results have already been context driven for several years now. The search results change depending on the context of the question. Context can mean your geographic location, previous search queries you have made or even the intent of your question (is the query informational or transactional). If your idea of SEO still includes fussing over keywords, you may wish to read this article I published in Search Engine Journal that discusses how modern search engines deal with keywords. This means that search results in the near future will be be context driven and less dependent on links.
Thumbnail overview of machine learning
Machine learning is about taking data and learning from that data. A simple example is examining search user click through and click back data to determine what kinds of sites tend to satisfy users for specific kinds of searches. Another example is creating classifiers of what kinds of content features are typical of high quality sites and using that to weed out content farms.
What is Allo?
Allo is Google’s new chat app that also includes an artificial intelligence assistant that eavesdrops on your chats (as well as listen to you) and can proactively suggest answers. It can answer queries within the context of a personal assistant. For example, if you tell it I want to eat a pizza, it can respond by providing you the phone numbers of local pizza restaurants.
When the AI gets an answer wrong it will use that information to learn from that mistake and thereby get better/smarter. A reporter for the New York Times test drove the Allo app and here is what the reporter observed:
To understand how Allo works, it’s easiest to think of the app’s A.I. assistant as an office intern who is lurking in the background, eager to chime in. The assistant analyzes messages you have typed or dictated and, when appropriate, springs into action with… suggestions for Google searches that may help accomplish tasks.
Google said it was still improving and refining its algorithms, and Allo’s assistant will get better over time.
Once Allo’s assistant matures, the Hangouts app will become redundant and you’ll be able to delete it from your device. The catch, of course, is that Allo’s A.I. won’t become sophisticated until more people use it and share feedback.
That last part about sharing feedback is important. It alludes to the millions of users who are creating a brand new data set for training the artificial intelligence within the context of a personal assistant.
Search is becoming a personal assistant
Scaled across millions of users, the feedback gained will enable Google to identify context on a general level (the difference between I want to make a pizza versus I want to eat a pizza) and on a micro level that is personalized just for you. This is an exponential leap away from the ten blue lines. It’s also a sign that Google and presumably Bing are making the leap away from links as a determinant of what is authoritative. Consider this, research from 2008 had already cracked the code for determining authoritative content without the use of links. That’s right. So instead of using links to determine authority, this method simply uses features on the web page itself (in this case, forums and blogs).
Does Google still use links for ranking?
Now imagine a scenario where Google has a clean set of data, an index of the web, that is more or less clean of on-page spam (Panda) as well as link spam (Penguin). At that point, Google might not have to use links to determine what content is the best content for a search engine user and in fact, this has been the case for many years now. Ever hear of the rank modification engine? It’s a way to modify the search engine results in order to better satisfy the user intent of a searcher, resulting in a set of web pages that are not necessarily ranked by which page has the most inbound links.
Allo is AI Search happening right under your nose
Now imagine a scenario where Google understands natural language queries given to a mobile app and can help a user obtain the information they need, outside the context of the ten blue lines. This is what Allo represents. As the New York Times reported stated, the more people use it the smarter it will become.
What makes it especially relevant to search marketing is that releasing an AI personal assistant is fantastic way of evolving information retrieval. A personal assistant version of search is the evolution of search to go beyond ten simple lines. If your livelihood depends on ten links in a search results page, then you better pay attention. Your world is about to be disrupted in a major way.
It’s not inconceivable to see this as a first step to changing the nature of search. In fact, I believe the nature of search has already changed. Ray Kurzweil has commented that humans tend to see progress as a linear process, like an evolution of incremental steps. He noted that technological change is exponential. With the release of Allo, the demise of a link based search marketing economy might be on the horizon. It has always been a matter of time.