Large Language Models (LLMs) provide strong summarization tools and can be used to summarize data on well-studied topics. However, they are not good at analyzing new topics that constantly emerge in our societies, which have been missing in their training data and may also be hard to collect real-time data for.

Additionally, LLMs fall short if the data available for a topic on the internet is biased, for example, when certain viewpoints are poorly represented online.

A related point is that LLMs are known to have a limited reasoning capability (See, for example, here). Some researchers attribute this limitation to the structure of neural networks, suggesting they are not inherently suited for reasoning. As mentioned earlier, this implies LLMs are not adept at developing new arguments from scratch.

Needless to say, LLMs also suffer from the hallucination problem, meaning they occasionally produce surprisingly incorrect results.

Despite these limitations, LLMs can, in some cases, generate a good set of initial arguments, which is valuable. Humans can then refine these arguments or reflect on the retrieved data to create more interesting topics.

Fact-checking is primarily achieved through the collective effort and knowledge of argument evaluators. If an argument lacks sufficient referencing, it is expected to repeatedly receive negative feedback during pairwise comparisons and be pushed to the bottom of the list.

The platform facilitates this process by making it mandatory for authors to select a Source Type when submitting their arguments. There are two source types to choose from: Self-explanatory and Linked References. The Self-explanatory type refers to cases where the argument is based merely on logical principles and does not require external references. In contrast, when the Linked References option is selected, the argument submitter recognizes that the claims being made in the argument require external references and presumably links the necessary references within the argument body. That is where the name Linked References comes from.

The requirement to provide a Source Type aims to nudge users to check if any references are needed to support their claims and, if so, to provide them.

Our answer consists of two parts. Firstly, it has been frequently observed in practical settings that the impact of misbehaving users on social platforms is often less significant than initially expected. We provide two examples. First, consider the online encyclopedia Wikipedia. When this platform was initially launched, many doubted that it could ever succeed given the risk of manipulation by ill-intentioned users. However, the platform is now one of the most visited websites on the internet. Second, consider the transportation company Uber. Many individuals again originally doubted that such a decentralized system could work due to the possibility of harassment by misbehaving drivers and passengers. However, the company ended up becoming very popular.

In the second part of our response, we emphasize our efforts at nlite to mitigate the impact of misbehaving users. One current area of focus is the development of an algorithm that helps detect the possible presence of two subgroups of users: one with good intentions that aims at ranking arguments in the proper order, and another with bad intentions that aims at ranking arguments either randomly or in reverse order.

It's important to note that if manipulative behavior is detected, the platform can publicize it, potentially damaging the perpetrators' reputation more than any early benefits they gain.

nlite can be a valuable tool in the following situations: (1) You are new to a topic and want to learn about it quickly and efficiently, or (2) You are already knowledgeable in a particular area and wish to share your expertise with others but lack an effective platform to do so. Both scenarios are explored in more detail below.

First, if you are new to a topic, nlite significantly increases the efficiency with which you can educate yourself about it. This is due to the neat structure of topic pages and the accuracy with which the top arguments are identified. Note that the alternative for you would be to spend numerous hours listening to debates or studying lengthy documents, and even then, you wouldn't know whether what you've come across are truly the best arguments that ever exist for the underlying viewpoints or if you've just learned about the insights of a limited number of experts.

Second, if you are knowledgeable about a topic—after all, we all have our own areas of expertise—and would like to enlighten society with your knowledge, nlite gives you a unique platform to achieve your goal. Note that once the top arguments for all sides are identified, it would be hard for the audience to look only at the arguments submitted for one side and ignore the others. Therefore, submitting quality arguments for the viewpoints you endorse will inevitably help promote them.