Not so long ago, we might have envisioned a futuristic world where artificial intelligence (AI) would perform a vast array of mundane duties. Is it possible?
For now, it would be unwise to underestimate the long-term potential of AI and its machine learning (ML) subcategory, or to assign it the omniscience it lacks. That aside, the potential for this technology is enormous, and the best way to describe the current situation would be to look at its “transitional state.”
Few things describe this mixed state of enthusiasm and suspicion better than the conversations and debates surrounding ChatGPT, a generative text model (or, more casually, a “chatbot”). For this reason, chatbots are an ideal starting point for a long-term discussion of generative AI that spans multiple blog posts, with the aim of examining its impact on intellectual property (IP) systems, their practitioners, and beneficiaries. increase.
Join us for a webinar to discover the world of cutting-edge legal technology and learn about the potential of artificial intelligence to revolutionize the trademark industry.
Register for the session on Thursday, May 25th at 5pm CET.
A Surprising Success: The Origins of ChatGPT
According to the creators of OpenAI (also the developer of the AI image generator DALL-E reviewed in this blog), ChatGPT was based on natural language processing (NLP), so no one is more surprised by its recent popularity. Apparently not. This model dates back to his 2020. OpenAI’s company leaders have approached development with the same mindset as the app’s development team, drawing on input from the tech industry with an updated NLP system that is more responsive and simulates conversational speech (and interactions) more effectively. Expected expectations. And hardly any mainstream attention.
And they were right—in the first place. ChatGPT received little attention when he opened to the public in November 2022, but by December he was making headlines in the news and on social media. According to early 2023 research, ChatGPT is the fastest growing consumer application of all time, with approximately 13 million monthly active users per day in January and over 100 million monthly active users. It is over.
The technical underpinnings of ChatGPT are not at all different from those of above-average e-commerce chatbots. As with the previous model, the AI text generator uses his NLP to interpret the user his request entered as typical speech (rather than a fragmented representation of a search engine query). But its sophistication and responsiveness are far superior to any customer service bot due to the vast amount of data it has been trained on.
- Ask him to write a script that combines fictional and real characters in outrageous situations. Realize it in a proper screenplay format.
- Asked why blues pioneer John Lee Hooker is so underrated, he details the poor sales and minimal promotion of his records. This oversight would only be made up for by his influence on later blues, R&B and rock artists.
- What about the issues most impacting the modern IP industry? ChatGPT quickly cites the pace of technological evolution, regulatory changes and cross-jurisdictional differences, IP trolling and counterfeiting.
ChatGPT’s utilitarian and humorous appeal is clear, but it still doesn’t seem to do enough to explain its lightning-fast user growth and Microsoft’s willingness to invest $10 billion in its marketing partnership with Bain. It seems. In fact, Coca-Cola’s next ad campaign will likely include slogans and content created by ChatGPT, as the soft drink giant is one of Bain’s most prominent customers.
Limitations and Controversies
OpenAI is very transparent about ChatGPT limits. Once logged in, users are presented with a disclaimer notice that historical knowledge beyond 2021 is limited (due to available training data), along with “occasionally incorrect information produced” and “harmful instructions”. A warning will appear saying that it may generate “biased content”.
Company leaders and programmers have been vocal about these issues in the media, claiming they are always working to address them. Major his update, announced on April 25, includes an option to turn off chat history to protect user privacy. This action will not be the last action taken in response to public complaints or government restrictions. New York City public schools recently banned the app over concerns of cheating, while Italy said it failed to verify users’ ages and that “there is no legal basis to justify the mass collection and storage of personal data.” Temporarily blocked access to the app. Following the Italian government’s reservations, regulators in Germany, France, Ireland and Spain are all monitoring the situation and taking action against OpenAI for possible General Data Protection Regulation (GDPR) violations. there is a possibility.
Perhaps surprisingly, the app’s ability to write code has generated the most heated controversy. It has been used to create malware (despite instructions from cybersecurity experts), although industry assessments have found it to be poorly programmed. Some in the industry feel they have the potential to replace human programmers, at least to some extent. Concerns that AI and automation are “job destroyers” are nothing new, and so far have not materialized as much as most of our most avid subscribers had hoped, but Goldman Sachs research predicts that generative AI will likely adversely affect 300 million offices. – Based jobs all over the world.
ChatGPT and IP
So far, most research into the actual and potential impact of generative AI on the intellectual property field has focused on image generation tools, the technology has sparked some scathing accusations, and copyright lawsuits have followed. never stop.
Nonetheless, as more lawmakers and copyright holders take notice of ChatGPT’s mechanisms and outcomes, this idiosyncratic focus on visual media is likely to end. But since chatbots are text-only, the potential for compromise is something of a gray area, if not non-existent. You can create material that is (intentionally) derived from the copyrighted work of an author, but it has not been legally verified whether the work created is up to the standard for infringement . Similarly, it has not been determined whether accessing and using copyrighted documents to train generative AI is for commercial purposes outside the permissible fair use/dealing in the first place. not. Presumably, the “education” exemption would not apply because AI is not a natural person.
Ultimately, these AI technologies, by their very nature to be generative, will have to somehow address key issues in the IP world, especially as they mature and expand their capabilities. The next blog in this series will tackle the issue of authorship and intellectual property ownership. stay tuned!
The content of this article is intended to provide a general guide on the subject. For your particular situation, you should seek professional advice.
