ChatGPT for Bing is almost here – is Google doomed?
Earlier this week we reported that ChatGPT is expected to have some kind of a launch on Bing, and indeed late Friday some sharp eyed Bing users spotted what appears to be an early version of ChatGPT integration into Bing, as noted by Owen Yin, one of those who had temporary access to “the new Bing:”
ChatGPT has seen a meteoric rise since it was first launched just a few short months ago in November of 2022, becoming the “fastest growing app of all time.” But is ChatGPT really ready to upend Search, to replace Google, and to become the next big thing? It may be just a bit early to declare all that. Let’s look at some reasons why.
Basic problems with bias and wrong information
As one of our readers notes in our post on the impending arrival of ChatGPT to Bing, the AI chatbot is already exhibiting signs of bias. It’s also prone to basic arithmetic errors, so while it’s surprisingly good at conversational text, the ability to generate human sounding answers to general topic questions, there’s still a long way to go before we can begin to trust ChatGPT.
Immediacy and learning model issues
Owen Yin says that “(u)nlike ChatGPT, which is trained on data collected up to 2021, the new Bing will be able to access current information.” I spent a bit of time “chatting’ with ChatGPT, and one of the questions I asked was if ChatGPT can train with live data. This was the response:
No, GPT-3 and other language models like it are not capable of training with live data in real-time. These models are trained offline on large datasets, and the training process can take several days or even weeks, depending on the size of the model and the amount of training data. Once the model is trained, it can generate responses in real-time based on the input it is given.
However, OpenAI and other researchers are exploring ways to fine-tune existing language models in real-time on smaller datasets. This process, known as online or continual learning, would allow language models to adapt to new data and improve their performance over time without having to retrain the entire model from scratch. However, this technology is still in the early stages of development and is not yet widely available.
So while Bing or other ChatGPT enabled search engines could readily help with questions like “how did the US enter World War 2?,” it’s going to continue to struggle with queries about, for example, that Chinese balloon that just got shot down a few hours ago.
Search monetization and the SEO industry questions
While ChatGPT “thinks” the issue of monetization (hint: Bing is not going to replace paid search with a free chatbot, no matter how cool it is) is within its realm. I asked if GPT-3 can be monetized:
Yes, GPT-3 can be monetized in various ways. OpenAI, the developer of GPT-3, offers API access to the model through its OpenAI GPT-3 platform, which allows businesses and developers to integrate the model into their applications and services. Other companies have built products and services on top of the API, such as chatbots, content creation tools, and virtual assistants, among others.
There’s also a whole SEO industry that’s probably a little bit nervous right now, but according to ChatGPT, there’s not much to worry about. I asked if ChatGPT can upend the SEO Industry:
It is unlikely that GPT-3 and other language models will completely upend the SEO industry. While GPT-3 has advanced capabilities for generating text and answering questions, it is primarily a language generation tool, not a search optimization tool.
Search Engine Optimization (SEO) is the process of optimizing websites to rank higher in search engine results and attract more organic traffic. This involves a number of techniques, including keyword research, on-page optimization, content creation, and link building, among others.
GPT-3 can assist with certain aspects of SEO, such as content creation and keyword research, by generating text that is optimized for search engines. However, it is unlikely to completely replace the need for human expertise in SEO, as it does not have the ability to perform technical SEO tasks, such as website auditing and link building, or to analyze and interpret search engine algorithms.
While GPT-3 and other language models may impact certain areas of the SEO industry, it is unlikely to completely upend it. SEO remains an important field that requires specialized knowledge and expertise to effectively optimize websites for search engines.
While the early look that some users got adds credence to the rumor that ChatGPT is coming to Bing soon, we’ll have to wait and see just how it’s implemented. Will it be widely deployed, or only run as a limited trial? Owen Yin notes that there will be a waitlist once “the new Bing” goes live, and we doubt it will be fully live for quite some time.
And what about that monetization model? Bing can’t be rushing to upend the very SEO industry it depends on, and will have to be very careful with how it messes with search monetization. Still, the idea of a new paradigm is exciting, and we’re looking forward to see what comes next.
Owen Yi’s post via Windows Central
Image generated using DALL-E 2
Technologie
via OnMSFT.com https://www.onmsft.com
February 4, 2023 at 11:40PM