OpenAI, the renowned artificial intelligence powerhouse, has unveiled its latest project, the web crawler named ‘GPTBot’. This move comes as a precursor to potential enhancements in future ChatGPT models and amidst rumours surrounding the development of GPT-5.
GPTBot: A New Web Crawling Tool
Similar to how search engines like Google and Bing operate, GPTBot functions as a web spider that indexes content across the vast landscape of the internet. OpenAI has signalled that the information collected using GPTBot could be a cornerstone in refining future ChatGPT iterations, potentially amplifying their accuracy and expanding their prowess.
OpenAI has provided clarity on GPTBot’s operations, stating in a recent blog post, “Web pages crawled with the GPTBot user agent may potentially be used to improve future models.” This implies that the information gathered could be instrumental in refining the precision and broadening the functionalities of subsequent ChatGPT versions.
However, OpenAI assures users that GPTBot’s data-collection will strictly centre around publicly available content. The web crawler will bypass sources demanding paywall access, those gathering personal data, and any text that contravenes OpenAI’s policies. Importantly, to respect web autonomy, site proprietors have the prerogative to block GPTBot by simply incorporating a “disallow” directive in a standard server file.
GPT-5: The Next Frontier?
The introduction of GPTBot closely trails OpenAI’s trademark filing for “GPT-5” at the United States Patent and Trademark Office on July 18. This trademark encompasses the utilization of GPT-5 for AI-powered human speech and text functions, as well as audio-to-text conversion and voice recognition features.
However, anticipation around GPT-5 might need to simmer a bit longer. OpenAI’s CEO, Sam Altman, in June, asserted that the firm is “nowhere close” to embarking on GPT-5’s training. He underscored the importance of exhaustive safety checks before initiating the project, suggesting a cautious approach to AI advancements.
Data Collection: A Thorny Issue
Recent months have seen OpenAI in the spotlight, facing scrutiny over its data acquisition methods, particularly regarding copyright and user consent. OpenAI encountered stern warnings from Japan’s privacy watchdog in June over unauthorized data collection. Similarly, April saw Italy temporarily barring ChatGPT, citing infractions against several European Union privacy directives.
Furthermore, a class-action lawsuit was instituted against OpenAI in late June, where 16 plaintiffs accused the AI behemoth of accessing private data from ChatGPT interactions. If verified, this could have serious implications for both OpenAI and Microsoft, the latter being named in the lawsuit. Such a breach could violate the Computer Fraud and Abuse Act, which holds precedents in web-scraping legal issues.
OpenAI’s launch of GPTBot signifies a proactive step towards enhancing AI capabilities, potentially paving the way for the eagerly awaited GPT-5. However, with ongoing legal challenges and privacy concerns, it’s evident that as the realm of AI expands, so does the scope of its ethical and legal considerations.
Source : Young, M. (2023, August 8). Circle CEO: 70% of USDC adoption comes from outside the US. Cointelegraph. https://cointelegraph.com/news/circle-ceo-usdc-adoption-emerging-developing-markets