Eps 1641: How ChatGPT will Affect Capitalism and You
— The too lazy to register an account podcast
| Host image: | StyleGAN neural net |
|---|---|
| Content creation: | GPT-3.5, |
Host
Ronnie Rodriguez
Podcast Content
You could even have ChatGPT explain a socialist critique of libertarianism using work of a specific socialist philosopher, such as G.A. If you want the most tame, canonical, politically infallible, high-level consensus on a given subject, just ask ChatGPT about it, and you will get three paragraphs of useful prose.
Which brings us back to college essays: a simple GPT-3 problem, as a lot of Twitter commenters claimed. At times, I would drop whole paragraphs of that essay in the ChatGPT, just to see if it would generate any good prose. I gave GPT-3 a question on the relation of language to color perception, which I had asked in a Psychology of Language course in my 3rd year, and it blew up.
GPT-3 may have failed on tests of reasoning, but it is excellent at producing human-like texts at varying lengths and styles. ChatGPT can write computer code, jokes , and dialogue for plays and TV shows. Compared with OpenAIs DALL-E 2, it seems that GPT-3 has better understanding of text input from users, as well as responses that it spews back.
With feedback from users, and the introduction of the more robust GPT-4 pattern, ChatGPT could be greatly improved in the future. ChatGPT has indeed reached new heights, which are ones the general community does not necessarily see as viable in this year. ChatGPT has generated much excitement, with some going so far as to speculate it may indicate a future where artificial intelligence has a dominant role over human content producers.
OpenAIs ChatGPT is not the first chatbot powered by artificial intelligence, and certainly not the last, but its intuitive UI and general efficiency leave the collective feeling that the future is coming. In the weeks since it launched, it is been reported that over one million users have given ChatGPT a spin, and that openAI is footing the bill. That is the way technology leaders are describing ChatGPT, a new chatbot model for conversations released by OpenAI last week.
The latest AI newcomer, ChatGPT is certainly the most impressive text-generation demonstration yet. ChatGPT is staggeringly impressive, impersonating a clever human uncannily using Generative AI, a piece of software that learns from massive sets of inputs in order to produce new outputs in response to users queries.
ChatGPT builds upon existing GPT natural-language technologies developed by openAI, the San Francisco-based organisation founded by technology leaders Sam Altman and Elon Musk, and supported by Microsoft, which helps to provide back-end cloud computing power to OpenAI products. ChatGPT is built by OpenAI, the start-up laboratory attempting to do nothing but create software capable of replicating the human mind. A tool like its latest creation can be used for real-world applications like digital marketing, creating online content, answering customer support inquiries, or, some users found, even helping to debug code.
ChatGPT has the potential to fill certain roles traditionally performed by humans, such as copywriting, answering customer service inquiries, writing press reports, and creating legal documents. As ChatGPT and other similar chatbots grow more widespread, they are likely to find applications in areas such as education and customer service. If history is any guide, the effects of tools like OpenAIs ChatGPT will largely be felt in existing industries, not by upending them via direct competition.
OpenAIs ChatGPT is not going to end universities; however, it may be the end of the college-essay-for-hire business. It is best to stop envisioning what tools like ChatGPT could achieve if they were deployed liberally and widely--as they are now, but will not forever, OpenAIs CEO Sam Altman has suggested--and start asking instead which potential uses would maximize revenues. Indeed, while vocal users insist that it is not all that clever, the ChatGPT model OpenAI made available to the public last week is so able to answer questions like humans that professionals in various industries are trying to digest the implications.
Just as any human coder can bring his or her biases to work, a machine generated by language, such as ChatGPT, holds the myriad biases found in the billions of texts that it uses to train its simulated understanding of language and thought. No one should mistake an imitation of human intelligence for the real thing, nor presume that the texts that ChatGPT regurgitates at the prompt are objective or authoritative.
When I asked ChatGPT to produce a speculative piece of capitalism-themed fiction, I got the distinct impression its heart was not truly in the task. For instance, we asked it to write a brief piece about Seattles startup ecosystem, and the results were relatively spot-on. Ethan Mollick, a professor of management at the University of Pennsylvania, had GPT-3 write a paper problem, generate a graded rubric for said problem, respond to it, and score their response.
That loop repeated itself last week, when AI firm OpenAI launched ChatGPT--a version of its text-writing bot that could apparently spit out any text, from Mozart-esque piano pieces to a Dr. Seuss-style London history.