ChatGPT reminds me of AOL
ChatGPT reminds me of AOL. Both acted as gateways to emerging technologies. AOL helped introduce many people to the internet, just as OpenAI, through ChatGPT, is making "artificial intelligence" widely accessible and user-friendly today.
In the 1990s, AOL was a visionary, bringing instant messaging, group chats, email, virtual gaming, digital news, file sharing, and web browsing to the general public. I was in college during this time and I remember the excitement of exploring cyberspace, logging into the information superhighway, and visiting the world wide web. Do you remember when people "surfed the web?" Or when you enjoyed hearing: "you've got mail!"
Similarly, ChatGPT is introducing groundbreaking applications to the mainstream, such as translation, transcription, image generation, data analysis, information synthesis, and content creation. It's akin to other technology introductions, like the Sony Walkman and iPod for portable music, Netflix for streaming movies, Blackberry for smart phones, the Ford Model T and Tesla for automobiles, and WordPerfect for word processing.
ChatGPT seems to be following a similar trajectory. With Microsoft's backing, OpenAI seems to echo Bill Gates' vision of a computer in every home, just as AOL sought to connect every house to the internet. ChatGPT aims to provide everyone with a personal virtual assistant.
Limitations
However, just as AOL showcased the limitations of dial-up, ChatGPT also reveals current technological boundaries. My experience with AOL was frustrating due to slow download speeds for music and videos, and frequent system crashes. ChatGPT4 has similar limitations. In my tests using the CNI Conference transcripts, the platform struggled with large files and often failed under complex demands, reminiscent of AOL's limited capacity.
I found that ChatGPT 4 would typically work on a task for three to four minutes. If it could not complete the operation in that amount of time, it just stopped and failed. I imagine future versions will be able to batch tasks. For example, work for three minutes (charge a token) -- pause -- and then resume (charging a second token.)
Despite these limitations, the potential for improvement is evident. Just as internet speed has dramatically increased since the dial-up days, it's likely that future versions of ChatGPT (and other general AI applications) will handle larger tasks more efficiently.
Beyond the Garden
As I thought more about this comparison, I remember leaving the walled garden of AOL. My roommates and I upgraded to broadband. This greatly increased our speed, but now all we had was a browser. The AOL interface and its portfolio of experiences were gone. We were in the wilderness now. The open uncensored web.
AOL, Prodigy, and CompuServe provided that initial step for many people. Not only getting them online, but curating an experience for them. That's what ChatGPT does today. It provides a friendly, easy-to-use, conversational interface for an initial "AI experience" by a mass audience: AI for non-programmers!
Before AOL, I dabbled with telnet, a command line environment. It was similar to using DOS before Windows. And I'm realizing that in order to genuinely explore the types of questions that I'm interested in -- ChatGPT is not powerful enough yet. Perhaps I’m not the target audience? For the large datasets that I want to create and the complex questions that I want to explore, I need to return to the command line by learning to use python and move things into Jupyter notebooks with Colab for compute power.
I'm thankful for ChatGPT because it showed me what's possible: I use it everyday. Computational tools like neural networks, machine learning, computer visioning, and so on have been interesting yet theoretical to me. GPT4, and mostly particularly the personal GPT utility, enabled me to dabble with emerging information science methods. It unlocked my thinking through pragmatic play, kinesthetically. It also reminded me that AI is more than ChatGPT, large language models, and generative AI. It's time to leave the garden again.
