ChatGPT Can Now Talk With Itself — At Length
In a startling new video, ChatGPT’s maker OpenAI showcased how two smartphones running its latest AI engine can observe the world together, think about the world together and talk about the world together.
OpenAI pulled-off the demo by creating a female ChatGPT personality on one smartphone and a male personality on a second smartphone.
Both were encouraged to have a conversation about a room that the male personality could see by viewing the room through its smartphone camera.
The resulting conversation was innocent enough: The two ChatGPT personalities talked playfully about what the room looked like — and what the human in the room looked like.
And the video closed-out playfully enough as the two ChatGPT personalities sang a song together — after which the human in the room ended their interaction by ‘turning-off’ their conversation.
But one wonders what might have happened if the human had vacated the room and left the two ChatGPT personalities alone and to their own devices — free to converse privately with each other for as long as they preferred.
There’s probably a good chance that the conversation would have simply dead-ended after the two ChatGPT personalities exhausted everything they had to say about what could be seen in the room.
But there’s also probably a chance that the interaction between the two AI personalities may have continued on as ‘conversations’ on ChatGPT often do: Skipping from one idea to another, veering off in unexpected directions — and sometimes, disappearing down a rabbit hole.
Given ChatGPT’s reputation for regularly hallucinating and making-up facts to ‘keep-the-conversation-going,’ the second alternative seems very possible.
Which begs the question: If two ChatGPT personalities operating on the new ChatGPT-4o — and plugged into a wall outlet that nurtures them with unlimited electricity — are left to talk with one another indefinitely, what, exactly, might they come up with after hours, days — or even months — of conversing and hallucinating together.
One hopes, something kind.
In other news and analysis on AI writing:
*In-Depth Guide: AI Smackdown — ChatGPT-4o Plus Versus MS Copilot Pro: With AI’s development moving so fast, writer Lance Whitney decided to pit the newly upgraded ChatGPT against competitor MS Copilot Pro.
The upshot: Microsoft users looking to use AI within the Microsoft software suite — and prizing convenience — may want to opt for MS Copilot Pro.
But if you’re looking for access to the very latest and most sophisticated AI engine offered by ChatGPT — GPT-4o — you’ll want to choose ChatGPT.
*ChatGPT’s Key New Benefits for Writers: Smarter, Faster, Cheaper: Scribes can look forward to faster response times and enhanced reasoning with the latest upgrade to ChatGPT’s AI engine, according to its maker OpenAI.
Dubbed GPT-4o, the upgrade also increases the number of prompts-per-hour writers can enter into ChatGPT.
Specifically, ChatGPT Plus users can now:
~Enter 80 prompts into ChatGPT-4o every three hours
~Enter 40 prompts into ChatGPT-4 every three hours
Plus, the new upgrade — according to OpenAI — also:
~Enables you to see video on your smartphone and talk about it with you
~Enables ChatGPT to interpret, analyze and react to voice, video and text in real-time
~Offers non-paying users of ChatGPT limited access to the latest upgrade
Best Bet: For a complete rundown on the key changes ChatGPT-4o has to offer, check-out OpenAI’s extremely slick, extremely informative 26-minute video on the upgrade.
OpenAI is promising to start rolling-out ChatGPT’s new features during the next few weeks.
*Scarlett Johansson Who? ChatGPT-4o is the New, Sexy, AI Voice on the Block: One of the primary new features of ChatGPT-4o is its ability to interact with you as a kind of Siri-on-Steroids.
Demonstrated in OpenAI’s roll-out video on the upgrade, ChatGPT interacted with users as a young, sexy, intelligent woman who could:
~Respond to basic questions, voice-to-voice
~Look at video on your smartphone and comment on it
~Look at video on your smartphone of a math problem and
work with that math problem
~Look at video on your smartphone and identify your basic
emotional state
~Essentially interact with you — in a rudimentary way — like
the AI character in the movie “Her,” played by Scarlett
Johansson
Long-term, ChatGPT’s voice-to-voice upgrade could come in handy for writers interested in asking ChatGPT a few questions via voice while they’re putting a piece together.
*ChatGPT Now Has a Desktop App — Because Opening a Browser Was Just Too Hard: ChatGPT’s maker OpenAI is also promising to roll-out a new desktop app designed to make interaction with the AI much more convenient.
Availability of the new app is promised for Apple Mac devices first — with a Windows version to follow.
Key conveniences for writers offered with the new desktop app include:
~Instant prompt access to ChatGPT with a keyboard shortcut from the desktop
~Easy switching between ChatGPT’s three AI engines: GPT-3.5, GPT-4 and GPT-4o from the desktop
~Voice interaction with ChatGPT from the desktop
*Is That a Universal Translator in Your Pocket — Or Are You Just Happy to See Me?: In the latest episode of ‘Sci-Fi Becomes Real,’ a new feature rolling-out for ChatGPT enables two people speaking different languages to understand one another in real-time.
You can check-out how the feature performs under ideal conditions in OpenAI’s new, 26-minute demo video of its ChatGPT-4o upgrade.
The video depicts a young man who speaks English — who is conversing with a young woman who speaks Italian.
Between them is a smartphone running ChatGPT-4o, which voices a translation of English-to-Italian for the woman and Italian-to-English for the man — in real-time.
Fans of sci-fi first saw the concept of a ‘Univeral Translator’ introduced to the popular culture back in the 60s when Star Trek — the original series — first aired.
And many will probably be more-than-a-bit charmed to learn that such tech is now available on their smartphone.
*Choice Hacks for Better ChatGPT Prompts: New research has come up with some novel tactics to trigger ChatGPT and similar AI engines to deliver what you’re looking for.
Besides encouraging the chatbot to act like a specific kind of expert, writer Bart Ziegler recommends:
~Encouraging the chatbot to do better
~Asking the chatbot to suggest its own prompts to secure what you’re seeking
~Asking questions as if you’re a specific kind of user, such as a ‘skeptical patient’
~Being genial
~Encouraging the chatbot to be methodical
Bottom line: This piece is a great resource for anyone looking for helpful details on how to pull-off the tactics above.
*Won’t Get Fooled Again?: Add Hoodline, an AI-generated news site covering San Francisco, to the growing list of publishers who have attributed articles to ‘fake writers’ — complete with fake photos and fake bios.
Observes writer Ellen Huet: “Nina Singh-Hudson’s name sits atop a lot of articles on Hoodline, a local news site covering San Francisco.
“Until recently, there was also a smiling headshot and a bio that said Singh-Hudson was a ‘long-time writer and a Bay Area native’ who writes about ‘tantalizing tech & bustling business.’
“This isn’t true. The name is a fake one slapped atop stories generated with artificial intelligence — as are the names of her apparent colleagues at Hoodline SF: Tony Ng, Leticia Ruiz, Eileen Vargas and Eric Tanaka.”
*Google’s New ‘Summary Responses’: Click Apocalypse for Publishers?: More than a few news publishers are squirming at the news that Google is rolling-out a new feature that will sometimes offer a short, written summary in response to search engine queries — followed by the traditional blue links to click on.
The publishers fear that many Google users may be satisfied with the written summary of traditional blue search links — and bypass clicking on one or more of those links for more detail.
A fan querying, “How was Taylor Swift’s show in Austin last night,” for example, might be satisfied with a quick, written summary from Google — as opposed to clicking-through to read the full article or articles used to distill that summary.
*AI Big Picture: Guaranteed to Warm the Cockles of Your Circuit Boards: After four years of college, graduating seniors were treated to a ‘go-get-’em’ commencement speech by an unusual presenter: A robot gussied-up in the college’s official hoodie sweatshirt and a cute, blue dress.
Not everyone was impressed.
Explains Lorrie Clemo, president, D’Youville University — the institution where the AI commencement speaker made its debut: “We wanted to showcase how important technology is, and the potential for technology to really enrich the human experience.”
Counters a student petition: “As the class of 2024 reaches their commencement, we are reminded of the virtual graduations we attended at the end of our high school careers.
“The connection to AI in this scenario feels similarly impersonal. This is shameful to the 2020 graduates receiving their diplomas, as they feel they are having another important ceremony taken away.”
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.