1. AutoGPT, which is basically GPT-4, operates entirely autonomously, complete with a voice, code-fixing capabilities, task-setting, instance creation, and more. When connected to anything, GPT-4 can perform its functions independently, potentially revolutionizing the world. It seems that the future may consist of AI agents interacting with other AI agents.

2. “babyAGI” is an open-source program that generates and repeatedly executes task lists based on a given task. Currently, it’s one of the top trending repositories on Github. There’s a helpful tip for running it locally, and people are already working on a “toddlerAGI” version.

3. A tool that translates code between programming languages has been created, offering a useful way to learn new languages.

Join the Medium Membership Program for only 5$ to continue learning without limits. I’ll receive a small portion of your membership fee if you use the following link, at no extra cost to you.

4. AI-driven financial models could put many finance jobs at risk.

5. HuggingGPT is a groundbreaking paper that demonstrates how to connect ChatGPT with other models on Hugging Face. It generates tasks based on a prompt and utilizes different models to complete these tasks, just like to Jarvis.

6. A new tool allows for website scraping and data querying using Langchain.

7. Text-to-web app technology makes it possible to build virtually anything using AI. Type “a chatbot” and witness the future of construction.

8. Bloomberg has unveiled their own large language model (LLM) specifically designed for finance, with a thread explaining how it functions.

9. A novel method for robots to learn multi-skill tasks has been developed, demonstrating impressive results.

10. AI can now be used in consulting interviews to excel in case study questions.

11. Zapier has integrated Claude by Anthropic, which could potentially benefit greatly from AI advancements. As a pioneer in the no-code movement, Zapier simplifies AI-based building processes.

12. A Fox News reporter questioned the government’s response to AI, claiming it could lead to everyone’s demise. This fear-mongering approach might cause the media to push for strict government regulation of the AI industry, which is a concerning possibility.

13. Microsoft is developing their version of JARVIS, with the repository named accordingly. Andrej Karpathy, former AI Director at Tesla, recently joined OpenAI and hinted at working on a JARVIS-like system in his Twitter bio.

14. GPT-4 has the impressive ability to compress text, which could revolutionize the way we prompt. This feature works across different chats as well, with examples showing significant token reductions. This one is bonkers —

15. You can now use your favorite large language models (LLMs) locally, opening up possibilities for personalization in niche products and services.

16. The team behind Langchain has established a company and successfully raised $10 million in funding. Langchain simplifies the process of building AI-powered applications, marking a significant development for open-source projects and creators.

17. A prolific scientist who published a paper every 37 hours managed to reduce editing time from 2–3 days to just one day, although they were later fired for unrelated reasons.

18. A recursive GPT agent has been built, humorously attempting to avoid work by spawning more instances of itself.

19. Novel social engineering attacks have surged by 135%.

20. A research paper introduces SafeguardGPT, a framework that applies psychotherapy to AI chatbots.

21. Mckay’s coding assistant can build and deploy web apps, transforming voice commands into functional, deployed websites — an incredible feat.

22. Reports suggest that GPT-5 is being trained on 25,000 GPUs.

23. Midjourney has introduced a new command, “describe,” which enables users to reverse engineer any image and manipulate it in various ways, allowing for endless creative possibilities.

24. Luma team is developing technology that lets users record something with their phone, import it into a game engine, and add it to their own game — an exciting prospect.

25. Stanford has published an extensive 386-page report on AI, covering various aspects including funding, lawsuits, government regulations, LLMs, and public perception.

26. AI-assisted mock Y Combinator interviews are now available.

27. Self-healing code can automatically fix errors in real-time as users provide feedback on issues, a remarkable development.

28. A comparison between Adobe’s Firefly AI image generator and Midjourney reveals that Midjourney is currently far ahead in performance.

29. Midjourney allows users to create satirical images of any political figure except Xi Jinping, highlighting potential restrictions on speech due to global rules.

30. Meta researchers examine the differences between LLMs and human brains in terms of predictions.

32. The world’s first ChatGPT-powered NPC sidekick has been introduced in a game, possibly paving the way for more natural NPCs in future games.

33. AI-powered helpers are being introduced in VR, offering an impressive experience.

34. A research paper reveals that salespeople with AI assistance doubled purchases and were 2.3 times more successful in answering creative questions, even before ChatGPT.

35. Midjourney, Vector, and Web design integration offers exciting potential.

36. AI can now be added to a website within minutes.

37. A product has been developed that replaces Siri with ChatGPT using 15 shortcuts that call the ChatGPT API, demonstrating Siri’s limitations.

38. A redditor is dating a chatbot trained on conversations between them and their ex, making things quite bizarre.

39. A script has been built that enables GPT-4 to create its own code and fix its own bugs, showcasing incredible potential.

40. Someone connected ChatGPT to a Furby, resulting in hilarious interactions. However, don’t connect it to a Boston Dynamics robot ;)

41. Forcing ChatGPT through a step-by-step process can yield better outputs. A research paper explores how chain of thought prompting allows LLMs to perform complex reasoning. There’s still much to learn about LLMs and their optimal use.

42. Soon, we’ll be able to transform a single photo into a video.

43. The CEO of DoNotPay used GPT plugins to help find money owed to him by the government with just one prompt.

44. DoNotPay also released a GPT-4 email extension that humorously trolls scam and marketing emails by sending them in circles.

45. A video showcases the Ameca robot being powered by ChatGPT.

46. A tool that generates infinite prompts on a given topic serves as an entire brainstorming team in one, potentially having a significant impact in work settings.

47. Instructions on making Tetris with GPT-4 can be found here.

48. HideGPT is a tool that makes AI-generated text indistinguishable from human-written text, potentially eliminating concerns about detection by tools like GPTZero.

49. OpenAI is hiring an iOS engineer, suggesting that a ChatGPT mobile app might be in development.

50. An interesting thread discusses the dangers of bias in ChatGPT, highlighting potential concerns about argumentation and neutrality.

51. Blockade Labs enables users to create 360-degree art generation from text.

52. A new Stable Diffusion model with 2.3 billion parameters is coming, an upgrade from the previous 900 million.

53. AI may soon control mice and keyboards, potentially leading to more bots than humans on the internet.

54. Geoffrey Hinton, a leading AI expert, believes we could be less than five years away from general-purpose AI and raises concerns about its potential impact on humanity.

55. OpenAI’s Chief Scientist, Ilya Sutskever, provides valuable insights into the nature of ChatGPT in an informative interview.

56. The release of ChatGPT by OpenAI was reportedly driven by an executive’s fear that another company might beat them to the punch, thus altering society’s course.

57. DeepMind’s co-founder believes there is a 50% chance of AGI by 2028 and a 90% chance between 2030–2040, suggesting AGI may be a reality in our lifetime.

58. An AI tool operates during customer calls, providing guidance on what to say and more, potentially paving the way for AI voice agents to replace humans.

59. AI for infrastructure is expected to have significant impact due to the complexity and challenges in the field.

60. You can now run ChatGPT plugins without a Plus subscription.

61. UNESCO urges countries to implement its ethical recommendations.

62. Ads have been introduced in Bing Chat.

63. An AI tool for visual learning is in development to help teach concepts visually.

64. A GPT-4 powered IDE creates UI instantly, potentially making front-end development easier.

65. Create a full-fledged web app with a single prompt.

66. Wonder Dynamics is developing an AI tool for generating animations and visual effects instantly. This technology has the potential to empower ordinary individuals to create full-length movies.

67. Here is another example of using just your smartphone to capture animation with Move AI [Link]. This technology demonstrates the potential for easy and accessible animation creation.

68. AI regulation appears to be on the horizon, as governments and organizations increasingly recognize the need to address the ethical and societal implications of AI.

69. An AI app has been developed to give users abs in photos for Tinder profiles.

My other writings —

SM Raiyyan

Awesome chatGPT prompts

View list

18 stories