ChatGPT created a lot of hype. Since the release of OpenAI’s LMM, there has been widespread speculation about how Generative AI (GenAI) will change everything about knowledge, research and content creation. How it transforms the workforce and the skills that employees aspire to. Or at the very least, how it would lead to a sudden boom in the industry as a whole. Here, they talk about search engines the most. Generative AI has the potential to dramatically change what consumers expect when they search. Despite the hype, generative AI presents practical, technical, and legal challenges that must be overcome before these tools become scalable, robust, and reliable.
Search engines have become popular since the early 1990s, but their basic approach remains the same: to offer users the web pages that are most relevant to them. The Search 1.0 era required individuals to enter a keyword or their combinations, and in this form, a question was asked for the search engine. Search 2.0 appeared already in the 2000s and introduced the process of semantic search, which allowed users to write natural phrases as if they were interacting with people.
And Google has become dominant for 3 reasons: it’s simple, it offers relevant answers, and it quickly finds websites that have the information you need. Now, it looks like another way to use it is emerging — since Google introduced Bard, users have been expecting more in-depth information. And that’s exactly what Search 3.0 does—it provides direct answers instead of web pages. In a figurative sense, Google is a colleague who points us to a specific book in the library, which, in turn, can answer our question. And ChatGPT represents an employee who has already read all the books in the library and can directly answer our question, at least in theory.
However, therein lies the first problem with ChatGPT: in its current form, it is not a search engine, mainly because it does not have access to real-time information in the way that web search engines offer. Interestingly, ChatGPT’s large database was studied, giving it static knowledge, along with its ability to understand and produce human language. However, apart from that, it does not know anything – for him, Russia has not yet invaded Ukraine, and the queen has not died.
Do you think this will change in the near future? The specific issue highlights a second, even bigger problem—at this stage, continuous training of LLMs is too much difficulty. This will require enormous energy and financial resources… but even if companies can overcome this technical or financial challenge, the problem that this time information suggests is still looming: What are tools like ChatGPT going to learn, and from whom?
Chatbots are like a mirror to society; they reflect what they see. Therefore, if you agree to train them with unfiltered data, you are guaranteed to make mistakes. Yes, that’s why they carefully select data for LLM. However, even this cannot ensure that all content from massive online data is factually correct and free from bias. According to the research, here we often deal with issues such as hegemonic views and biases that exist against marginalized communities.
This is a big problem because it drives users to the very websites that offer biased, racist, wrong, or otherwise inappropriate content. But here Google is just a guide that points people to sources and takes less responsibility for their content. Users themselves must distinguish between facts and objective opinions and decide which information to use. ChatGPT does not have the latter, therefore, it is not biased toward any particular viewpoint or agenda. However, as an AI language model, ChatGPT has been trained on vast amounts of data from the internet, which means it may have inadvertently picked up biases that exist within that data. It is important to acknowledge this and work towards minimizing any biases that may exist.
One of the challenges with online content is the ease with which false information can spread. This can be particularly harmful when it comes to issues such as public health or politics, where inaccurate information can have serious consequences. Fact-checking websites and services are essential to combating the spread of misinformation online.
Another issue is the echo chamber effect, where users are only exposed to content that reinforces their existing beliefs and opinions. This can lead to a lack of exposure to diverse viewpoints and can exacerbate polarization. Google’s search algorithm has been criticized for contributing to this phenomenon, as it tends to prioritize content that is similar to a user’s previous searches and clicks.
In recent years, there has been increased attention on the responsibility of tech companies to moderate the content on their platforms. Social media companies such as Facebook and Twitter have faced criticism for their handling of hate speech and disinformation. Google has also been called upon to do more to ensure that its search results are not contributing to the spread of harmful content.
Ultimately, it is up to both tech companies and individual users to take responsibility for the content that is consumed and shared online. While algorithms and search engines can guide users towards information, it is important to always be critical of the sources and to seek out diverse perspectives. By doing so, we can work towards a more informed and tolerant online community.
