User add493145479... - AI Fiction in the Wild

IP #223

Back to search

1 fiction conversations over 5 days (Feb 11, 2024 – Feb 17, 2024) 66% prompt uniqueness North Yorkshire, United Kingdom

Possibly Same User

These IP addresses have similar average prompt embeddings (cosine similarity ≥ 0.85, same state). View as estimated user →

578f71de5f0d... ( conv.) 92f5a7cf0a2e... ( conv.) b4b912a6e7f4... ( conv.)

Activity Timeline ▶

Feb 11, 2024

Feb 17, 2024

All conversations

Clustering: TF-IDF Sentence DBSCAN Per-user

Conversation Clusters (78)

▼

1 conversation in cluster Cluster 65× “What Would 1981, 1985, 1987, 1990, And 1993 Versions of LaMD...” over less than a minute (Feb 14, 2024 – Feb 14, 2024), oldest first. Show: 10 15 50 100

Feb 14, 2024 at 10:01 AM GMT — 2 turns United Kingdom, North Yorkshire fiction View full conversation

USER: What Would 1981, 1985, 1987, 1990, And 1993 Versions of LaMDA Look Like?

CHATBOT: LaMDA, Google's conversational AI, did not exist in any of those years, as it was only announced in 2021. However, for the sake of a creative thought experiment, let's imagine what an equivalent level of technology might have looked like in 1981, 1985, 1987, 1990, and 1993 if somehow the concept and underlying algorithms could have been developed that early: **1981:** - Computing power and storage were very limited in 1981. The concept of a language model like LaMDA would be beyond the scope of the technology at the time. However, if it somehow existed, it would likely be a massive, room-sized mainframe computer with access to a relatively small database of text, capable of basic pattern recognition and responses. The interface would be text-based, likely accessed via a command-line interface on a terminal with a monochrome display. **1985:** - Still before the World Wide Web, a 1985 version of LaMDA would likely involve a slightly smaller mainframe or several interconnected minicomputers. Personal computers were becoming more prevalent, but they lacked the capacity to run advanced natural language processing algorithms. Imaginary LaMDA might use simple rule-based systems for text interactions and be available to a select few researchers or large corporations. **1987:** - By 1987, computers had become more powerful, and AI research was experiencing a renaissance. However, any version of LaMDA would still be primitive and constrained to academia or industrial research labs. It might use early forms of neural networking and have increased its text corpus size significantly, but it would still be far from real-time natural language understanding. **1990:** - The early '90s saw the rise of the internet and more advanced consumer hardware. An early LaMDA could have been a distributed system across several workstations, allowing for more processing power and storage. It would likely still be rudimentary and used primarily for research, but with the advent of the internet, the potential for accessing larger datasets would make it more sophisticated in understanding and generating text. **1993:** - The tech world was on the cusp of major advancements in networking and the World Wide Web. A 1993 LaMDA-esque system might be accessible to more users online through a primitive web interface or bulletin board systems (BBS). Still, it would likely offer slow and limited conversational capabilities, with responses that would seem canned and very obviously computer-genera...

USER: What Would The Hypothetical 1981, 1985, 1987, 1990, And 1993 Versions of LaMDA Respond to the Question "What Was Michael Jackson's First Hit Single?

CHATBOT: Given the limitations of theoretical technology at those times, let's speculate on the kind of responses you might get from these anachronistic versions of LaMDA to the question "What was Michael Jackson's first hit single?": **1981:** - With a limited database that might not even include real-time access to information, a primitive AI system might respond with a very generic answer or it might not know the answer at all. If well-programmed, it could potentially retrieve Michael Jackson's earliest known songs from a static dataset, if the information was pre-loaded. For instance, "Got to Be There" from 1971 could be a response, but the system might not be able to distinguish between his "first" single and his "first hit" single. **1985:** - A slightly more advanced AI might have access to a larger dataset, perhaps including some music information. However, given the lack of real-time internet connectivity, the information would still be static and potentially outdated. The system might be able to list a few of Michael Jackson's early songs but could struggle with context. It might say something like "Michael Jackson's early successful singles include 'Got to Be There' and 'Ben.'" **1987:** - An improved system might now understand basic context and have a larger dataset to pull from, including more detailed music history information, maybe even pulling from CD-ROM encyclopedias. The response could still be inaccurate but slightly more informed, such as "Michael Jackson began his solo career in the early 70s. 'Ben,' released in 1972, was among his first hits as a solo artist." **1990:** - By 1990, the system might be connected to a nascent internet with access to some newsgroups or databases with up-to-date information. This could allow it to provide a more timely answer, like "Michael Jackson's first major hit single as a solo artist was 'Off the Wall' from the album of the same name, released in 1979." The system, however, might still struggle to recognize that "Off the Wall" was not actually his first hit solo single (that honor goes to "Got to Be There" or "Ben" from the early '70s). **1993:** - With access to early online databases or information services (such as those provided by CompuServe or America Online), a system in 1993 could theoretically search for more current information. Its response could be more accurate, something like, "Michael Jackson's first hit solo single was 'Got to Be There,' released in 1971. He gained fame as a member of the J...

Cluster 65