Most of us don't read these terms and conditions when shopping online. Most people believe that the terms in these standardized agreements are non-negotiable and hope that they serve the best interests of the majority.
However, this doesn't seem to be the case in many situations. From laptop manufacturers to airlines to companies that offer buy now, pay later options, there is endless debate about whether the terms and conditions are fair.
It is difficult to obtain exact facts about the scale of this issue, but there are indeed many dissatisfied shoppers on consumer forums.
For example, many people encounter difficulties in exercising their contractual rights or purchasing items because they are unaware of the aspects they agreed to in the contract. In such cases, it is difficult to find free or affordable legal advice, so many people may give up.
While people should read the contract before clicking "I agree," most people may not do so due to time and ability constraints. However, there may soon be a solution. We will soon be able to use artificial intelligence to read all the fine print without having to delve into it ourselves.
Existing Solutions
Artificial intelligence tools for analyzing legal documents have been around for some time. They can identify potential issues, such as rights violations that consumers may need to further investigate. However, due to the limitations of processing large amounts of text, AI tools must copy and paste the terms sentence by sentence. Additionally, such tools are designed to read self-explanatory guides rather than completely eliminate the need to read the terms.
More sophisticated AI tools address the issue of reading website policy documents. They do not require text copying and pasting but instead analyze the relevant URLs. The focus here is on how internet service providers use your data. This makes it easier to teach the AI models everything they need to know, especially in such a heavily regulated field.
The challenge with terms and conditions lies in their diversity. Suppliers have more freedom in their wording, making it more difficult for AI to detect and understand these terms and conditions.
There are also many differences between jurisdictions. For example, in the UK, it is "solicitors," while in the US, it is "attorneys," which means AI trained on US data may mislead UK consumers. However, existing tools often do not explicitly indicate which jurisdiction they are designed for.
You may wonder if you can copy and paste the terms and conditions into the latest AI chatbots, such as ChatGPT, but this is not a solution. These general models are not specifically trained for legal texts or legal analysis. This means that any advice they provide may be accurate, inaccurate, or entirely fictional.
Addressing the Issue
According to our understanding, there are no teams of developers currently attempting to create dedicated AI models, such as OpenAI's GPT-4, for creating consumer-specific terms and conditions. ChatGPT is built on top of GPT-4.
Instead, many AI developers seem to be focused on creating tools to automate legal work for law firms and other companies. This may even lead to conditions that are less favorable to consumers, as the focus is likely on cost-cutting rather than improving service quality.
To change this situation, the main author of the research, Jens Krebs, and his colleague Ella Haig, have been developing a terms and conditions application for England and Wales at the University of Portsmouth. Once fully developed, people will be able to copy and paste the entire document into the application for analysis.
The application will then list any terms that may not comply with consumer legislation, such as the Consumer Rights Act 2015, which may have unexpected effects on consumers. It will also compare all the terms with similar terms used by other suppliers to ensure there are no abnormal terms. When it identifies anomalies, it will advise consumers to read that section before deciding whether to proceed.
Currently, the project is testing different AI models for the application to determine which one is most effective. So far, Google's BERT model has performed the best, with an accuracy rate of 81%. Researchers are testing the application with test data to validate its accuracy, where they know what the correct results should be.
The project will not release any programs until the accuracy reaches 90% to 95%. The aim is to release the program to consumer organizations like Which in 2024 and have a full release in 2025. The plan is to make the program free to use.
One key obstacle that such projects need to overcome is the lack of examples of harmful terms, making it difficult to provide enough data for AI training. This is the problem consumers face when trying to assess terms and conditions. The long-term plan for the Portsmouth application is to collect real data from consumer organizations, governments, and consumers to supplement and replace the training data of the application, thereby continuously improving its accuracy.
People hope that this application will lead the way for a new generation of AI tools aimed at making terms and conditions more transparent. In addition to potentially reducing the number of dissatisfied consumers, these tools can also help those who have already signed unreasonable terms to prepare and present their own cases, reducing the need for lawyers.
If such services become popular, they are also expected to discourage suppliers from pushing acceptable boundaries. If terms and conditions become more favorable to consumers, it will be a significant victory for this emerging technology.
Source: https://techxplore.com/news/2023-06-ai-consumers-unfair-terms-conditions.html
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.