Imagine a world without Google, without those pesky Ads that ruin the search experience for billions of users who are lucky enough to have access to the internet.
GPT-3, One of OpenAI’s most renowned language models, is being used by more than 300 applications for searches, conversations, text completions, and other AI features through OpenAI’s API that was launched just 10 months ago. It’s entirely possible that future iterations of GPT lead to more advanced capabilities in search than are possible today in 2021.
Amazon has already disrupted Google in product search, Google is spamming Ads at the top of every first page, YouTube is going wild with Ads, but all of these advertisements really lowers the quality of the internet. GPT-3’s language abilities are improving at such a pace that this technology could realistically disrupt Google, Bing and all the others.
With tens and thousands of developers building on the platform globally, OpenAI is currently responsible for generating an average of 4.5 billion words per day. The API has been designed in such a simple and flexible way that is making the machine learning teams more productive, without having to put much effort.
GPT-3 is credited with language and content disruption online, but what if it disrupted the fundamental human interface, search? Would such a reason mean Microsoft will be tempted to acquire OpenAI sooner than expected?
AI will Disrupt the Ad-Driven Internet and GPT Could Do It
We know computers are getting better at writing, but eventually this will mean they are getting better at search. Search and writing are so fundamental to how humans act on the internet but what if AI is able to do it better and faster? What kind of an internet does that lead to and how could it disrupt the ad-centric internet as we know it? With AI doing the act of search and content generation, perhaps advertising would no longer be as effective with less of those ‘eyeballs’ scanning screens. What would be the point of an Ad on a search page if that reality comes into being?
GPT-3 also has a free alternative, GPT-Neo. How much innovation will come from OpenAI’s spawning GPT? Google and Facebook hijack our attention and therefore have incentives to place Ads there. But what if in the near future that process was disrupted? The way we search online hasn’t changed in decades but eventually we’ll be talking to AI like talking to expert and we won’t tolerate Ads getting in the way.
AI Language models will easily replace page rank and page ranking algos. BERT or GPT-3—or a future version of them will likely do this, and it very well might not be done with Google. Google has too much at stake to do it right, without Ads.
It’s biased due to its profit-motive, it doesn’t really care about the user experience, it’s a monopoly. So it’s better if OpenAI or someone else does it, creates the next Google. It needs to have a monetization model that disrupts Ads and junk spam on screens.
How Google will be Disrupted
It’s very simple really, AI will lead to a better Google that’s not like Google that improves the user experience of search and the AI-human interface. Language really is the key, and I think at this point OpenAI understand this. Search engines have become faster and more accurate, even as the web has exploded in size.
AI is now used to rank results, and Google uses BERT to understand search queries better. However what if there was a competitor that understood how to fuse search and language models better than Google? Surely that’s inevitably going to happen. Imagine how valuable such a company would be?
The problem is that even the best search engines today still respond with a list of documents that include the information asked for, not with the information itself. With better language models personalized to each user, how we search online could be disrupted in the next five to ten years, circa 2029.
Search engines are also not good at responding to queries that require answers drawn from multiple sources. Search is still dumb and the interface is still dumb for Google to spam its Ads at us. But that’s invasive, inefficient and very 1999. Google has kept things dumb on purpose, for profit.
Let Amazon create a real search engine based on language models and a new human-AI interface, heck let ByteDance do it. If Americans can’t get this right, surely the Chinese will. Today, GPT-3 draws information from multiple sources to answer questions in natural language.
The problem is that it does not keep track of those sources and cannot provide evidence for its answers. However in future iterations of GPT, these limitations could be corrected. Talking to AI in the future will feel more like talking to an expert than talking to an algo whose main purpose is to get you to click on Ads.