Is it possible for ChatGPT to write a screenplay? If so, would it actually be good? I couldn’t help but wonder, so I decided to run some tests and am sharing them here today.
By now you’ve all heard about ChatGPT, and know how it can answer just about any question. But perhaps its greatest strength its ability to write and edit huge amounts of text almost instantaneously.
Naturally, I couldn’t help but wonder what would happen if ChatGPT were tasked with developing a feature length screenplay idea…
So I tried it out – Not because I want to outsource my writing (that would completely defeat the purpose for me), but to get into the “mind” of AI and see how it approaches the writing process. Read More
Monthly Archives: May 2023
Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learning
Soccer players can tackle, get up, kick and chase a ball in one seamless motion. How could robots master these agile motor skills?
We investigated the application of Deep Reinforcement Learning (Deep RL) for low-cost, miniature humanoid hardware in a dynamic environment, showing the method can synthesize sophisticated and safe movement skills making up complex behavioral strategies in a simplified one-versus-one (1v1) soccer game. Read More
Large Language Models and Elections
Earlier this week, the Republican National Committee released a video that it claims was “built entirely with AI imagery.” The content of the ad isn’t especially novel—a dystopian vision of America under a second term with President Joe Biden—but the deliberate emphasis on the technology used to create it stands out: It’s a “Daisy” moment for the 2020s.
We should expect more of this kind of thing. The applications of AI to political advertising have not escaped campaigners, who are already “pressure testing” possible uses for the technology. In the 2024 presidential election campaign, you can bank on the appearance of AI-generated personalized fundraising emails, text messages from chatbots urging you to vote, and maybe even some deepfaked campaign avatars. Future candidates could use chatbots trained on data representing their views and personalities to approximate the act of directly connecting with people. Think of it like a whistle-stop tour with an appearance in every living room. Previous technological revolutions—railroad, radio, television, and the World Wide Web—transformed how candidates connect to their constituents, and we should expect the same from generative AI. This isn’t science fiction: The era of AI chatbots standing in as avatars for real, individual people has already begun, as the journalist Casey Newton made clear in a 2016 feature about a woman who used thousands of text messages to create a chatbot replica of her best friend after he died. Read More
There Is No Turning Back on AI
The only question is: will we learn how to live in moving history?
ChatGPT has been around for about two seconds and it’s already changing how people work, how they write, how they research, how they cheat on tests, how they profess their love, and what they make for dinner.
We didn’t have to talk about theoretical “use cases,” as people still do with cryptocurrency. ChatGPT, which 100 million people use every day, showed us its uses right away.
That is really exciting. And also really unnerving. Read More
Are Emergent Abilities of Large Language Models a Mirage?
Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models. What makes emergent abilities intriguing is two-fold: their sharpness, transitioning seemingly instantaneously from not present to present, and their unpredictability, appearing at seemingly unforeseeable model scales. Here, we present an alternative explanation for emergent abilities: that for a particular task and model family, when analyzing fixed model outputs, one can choose a metric which leads to the inference of an emergent ability or another metric which does not. Thus, our alternative suggests that existing claims of emergent abilities are creations of the researcher’s analyses, not fundamental changes in model behavior on specific tasks with scale. We present our explanation in a simple mathematical model, then test it in three complementary ways: we (1) make, test and confirm three predictions on the effect of metric choice using the InstructGPT/GPT-3 family on tasks with claimed emergent abilities, (2) make, test and confirm two predictions about metric choices in a meta-analysis of emergent abilities on BIG-Bench; and (3) show how similar metric decisions suggest apparent emergent abilities on vision tasks in diverse deep network architectures (convolutional, autoencoder, transformers). In all three analyses, we find strong supporting evidence that emergent abilities may not be a fundamental property of scaling AI models. Read More
Midjourney 5.1 Arrives – And It’s Another Leap Forward For AI Art
Midjourney 5.1 has been released, bringing another significant improvement in the quality of results from the generative AI art service.
The company claims that version 5.1 of the engine is “more opinionated”, bringing it closer to the kind of results that you would get with version 4 of Midjourney, but at a higher quality. There’s also a “raw” mode, for those who don’t want images that are as strongly opinionated.
Other claimed improvements include greater accuracy, fewer unwanted borders or text artifacts in images, and improved sharpness. Read More
Meet Pi, a new ChatGPT-like AI for personal assistance and emotional support
Pi has been launched by Inflection AI, a startup founded by LinkedIn co-founder Reid Hoffman and Google DeepMind co-founder Mustafa Suleyman.
“Hey there, great to meet you. I’m Pi, your personal AI.”
This is the message one is met with when you open Inflection AI’s chatbot, which is the latest addition to a stream of chatbots introduced by big tech companies over the last couple of months.
Pi, it’s a ChatGPT-like competitor designed to be a kind and supportive companion assistant. Read More
Chatbot ‘journalists’ found running almost 50 AI-generated content farms
Chatbots pretending to be journalists have been discovered running almost 50 AI-generated “content farms” so far, according to an investigation by the anti-misinformation outfit NewsGuard.
The websites churn out content relating to politics, health, environment, finance and technology at a “high volume”, the researchers found, to provide rapid turnover of material to saturate with adverts for profit. Read More
Futuri Launches RadioGPT™, The World’s First AI-Driven Localized Radio Content
Live, local, and powered by AI, RadioGPT™ marries GPT-3 technology with Futuri’s TopicPulse content discovery AI and voices generated by artificial intelligence to create the world’s first 100% AI-driven radio hosts.
Cleveland, Ohio, February 23, 2023 — Futuri is revolutionizing the audio industry with the launch of RadioGPT™ — the world’s first AI-driven localized radio content solution. RadioGPT™ combines the power of GPT-3 technology with Futuri’s AI-driven targeted story discovery and social content system, TopicPulse, as well as AI voice tech to provide an unmatched localized radio experience for any market, any format
RadioGPT™ uses TopicPulse technology, which scans Facebook, Twitter, Instagram, and 250k+ other sources of news and information, to identify which topics are trending in a local market. Then, using GPT-3 technology, RadioGPT™ creates a script for on-air use, and AI voices turn that script into compelling audio. Read More
OpenLLaMA: An Open Reproduction of LLaMA
In this repo, we release a permissively licensed open source reproduction of Meta AI’s LLaMA large language model. In this release, we’re releasing a public preview of the 7B OpenLLaMA model that has been trained with 200 billion tokens. We provide PyTorch and Jax weights of pre-trained OpenLLaMA models, as well as evaluation results and comparison against the original LLaMA models. Stay tuned for our updates. Read More
Rick's Cafe AI 