The future of conversation AIs

When Microsoft first released OpenAI-powered Bing everyone was so surprised. It is still in beta but it has already started a new era of web search and information delivery, thanks to OpenAI. After using ChatGPT, we already knew how capable conversation AI would be if it had access to current information and Bing delivered precisely that. It seems natural and already kind of makes Google search feels obsolete. Google quickly released a similar LLM-powered conversation AI - BARD and it is safe to say now that the response from the media is not great. BARD feels much more constrained compared to Bing. Many people even say that Google is lagging behind Microsoft now in the AI race.?

But I don't think that is true. I think Google is playing it safe and still trying to figure out how to tackle this ChatGPT problem, and how to integrate BARD nicely with its behemoth advertisement business. Remember that Google is first and foremost an advertisement company, In Q4 2022, Google’s revenue totaled $76 billion, of which $42.6 billion (or 56%) was from search ads. Revenue from Google Network ads hit $8.5 billion (11.1%), and YouTube ad revenues registered at $8 billion (10.5%). So search advertisement is a pretty big deal for Google, unlike Microsoft which has a much more diversified portfolio and revenue streams.?The real threat of ChatGPT-like conversation AI is that a user can get answers directly from ChatGPT without leaving that chat page, which is a problem for Google as it serves advertisements against a search query hoping that user will click that sponsored link to find more. But in the context of a conversation AI it is more difficult to do it right as AI itself will serve and summarize the answers, with no reason to visit links in many cases --- meaning less advertisement money. Bing is trying to tackle this problem by sharing information source links with its answers and it also started to serve relevant ads against a search query along with answers but there are still many unanswered questions and a lack of revenue performance data. Microsoft can take this kind of risk with Bing as it is not its core business, but Google can't. That's why Google is moving slowly, not because they are lagging in AI research, but because of questions related to revenue.

Also, LLM-powered conversation AIs raise many questions. How content creators like publishing websites can make money if AI delivers all the content directly without sending users to their websites? Loss of traffic means fewer ad impressions and less money. These AI engines are trained on the data scraped from the websites and there is no framework or government regulation now to enforce these AI engines to share revenue with publishers. One notable example can be StackOverflow. ChatGPT or Bing can now solve concerning amount programming-related questions directly, and they are most probably trained on StackOverflow data. Previously users need to search in a search engine and the search engine could point them to relevant StackOverflow questions where they can get the answers, now ChatGPT/Bing is doing it directly without ever leaving that chat window. StackOverflow can be a serious victim here as it will see less traffic later on, which means less interaction with the questions posted there --- leading to decreased community activity. Reddit, Quora, or any similar sites are facing real danger from these conversation AIs. Revenue sharing is a big unanswered question here. Similar things can be said for code generators like Github Copilot. GDPR compliance in Europe is another thing that should be a concern for AI companies and can be a hard one to solve.?

These conversation AIs are a big revolutionary step toward the future and the future is an exciting one. These AI engines can truly transform how we work and learn every day, but mass adaption I think is still quite far away and can create a huge chaos in the creator community without proper content monetization and strict policy enforcement.?