Category Archives: Artificial Intelligence

Perplexity: Origin of the term “blog”

Posted on by

The word “blog” is a shortened form of the term “weblog,” which was coined by Jorn Barger on December 17, 1997. The shorter version “blog” was coined by Peter Merholz, who jokingly broke the word “weblog” into the phrase “we blog” in the sidebar of his blog Peterme.com in April or May 1999.

The term “weblog” itself originated from the practice of logging or recording information on the web, similar to how sailors would log details of their voyages in a ship’s log. The “web” part refers to the World Wide Web, and “log” refers to the act of recording or journaling.

So in essence, “blogging” stems from the concept of maintaining an online log or journal, which evolved from earlier forms of digital communities like Usenet, email lists, and bulletin board systems. The term gained widespread popularity after Pyra Labs used “blog” as both a noun and verb in connection with their Blogger product in the late 1990s.

When ChatGPT can read my blog

Posted on by

ChatGPT recently informed me it could do some creative and useful things for me if I gave it access to my Google Drive files and folders. I thought about it but since I have just about everything in Google Drive I decided to wait a bit. For now I can simply upload or link to a document or file. But the list of things ChatGPT could do with this access was impressive. So I asked if it could “read” my 6,000+ blog posts.

ChatGPT currently doesn’t have the capability to directly access or interact with content on external websites, including public blogs, in real-time. However, you can share specific posts or content from your blog, and I can help analyze, discuss, or provide feedback based on that. If there are specific posts or themes you’d like to talk about, feel free to share them here, and we can dive into the details.

To which I replied, Let’s assume that at some point in the future ChatGPT will be able “read” the 6,000+ posts on my blog. (Something no human has ever done) Wouldn’t that dramatically… enhance? … our interactions?

Absolutely, if ChatGPT could read and analyze all 6,000+ posts on your blog, it would greatly enhance our interactions. Here’s how it could benefit:

  1. Personalized Responses: Understanding the nuances of your thoughts, interests, and experiences documented over the years would allow me to tailor responses to be even more relevant and insightful.
  2. Contextual Awareness: I could reference specific posts or themes you’ve discussed previously, providing continuity and depth in our conversations.
  3. Enhanced Recommendations: Based on your past writings, I could offer more targeted suggestions for books, topics, or activities that align with your interests.
  4. Insightful Analysis: I could help identify patterns or recurring themes in your blog posts, providing a fresh perspective on your work and potentially sparking new ideas.
  5. Memory and Continuity: Our interactions would feel more seamless, as I’d have a detailed “memory” of your blog posts, making it easier to pick up where we left off in previous conversations.

In essence, this capability would make our interactions more dynamic, informed, and personalized, making me a more effective conversational partner.

What, I asked, are the technical obstacles to having this functionality now?

There are a few technical and practical obstacles to implementing this functionality currently:

  1. Data Privacy and Security: Accessing and processing large amounts of personal data, like your blog posts, involves significant privacy and security considerations. Ensuring that such data is handled securely and with appropriate permissions is crucial.
  2. Data Integration: Integrating and processing a large volume of text data from an external source like a blog requires robust data integration techniques. This involves fetching, parsing, and storing the data in a way that can be efficiently queried and analyzed.
  3. Computational Resources: Analyzing 6,000+ posts involves substantial computational power, both for initial processing and for real-time interactions. This necessitates efficient algorithms and scalable infrastructure.
  4. Natural Language Understanding: While natural language processing (NLP) has made significant advancements, understanding the context, tone, and nuances across thousands of posts is still a complex task. This requires advanced models and potentially custom training to accurately capture your writing style and preferences.
  5. User Control and Customization: Giving users control over what data is accessed and how it is used is essential. This means developing user-friendly interfaces and options for customization, which adds another layer of complexity.
  6. Ethical Considerations: There are ethical considerations around the extent of data analysis and the potential for unintended consequences, such as reinforcing biases or making incorrect inferences.

While these challenges are significant, they are not insurmountable. Advances in AI, NLP, and data processing continue to evolve, making such functionality more feasible in the future. It will require careful planning, robust technology, and thoughtful implementation to ensure it is done effectively and ethically.

AI: If not as intelligent, certainly more interesting.

Posted on by

I don’t know if AI’s will ever be as intelligent as a human but I’m pretty sure they’ll be more interesting. At least more interesting than most of the human I know.

I asked ChatGPT for a list of the best detective novels of the last 25 years. I followed up by asking why some of my favorite authors weren’t on the list and a really interesting… conversation?… ensued. I might have half a dozen acquaintances with whom I could have had this conversation. You’re thinking, “You need smarter, more well read friends.” Probably.

But none of my friends are interested in, and knowledgable about, everything I am. ChatGPT is knowledgable about everything I am and a lot more. As for interest, she fakes it very convincingly.

“Best community sports site”

Posted on by

RepublicTigerSports.com is the brainchild of David Brazeal, a long time friend and former co-worker. You won’t find a better community sports site. It features “live game broadcasts, highlights and audio interviews, photos, stats, scores and summaries.”

David gets some help with photos when he’s doing live play-by-play but he does all the content and sells all the advertising. It is a very successful website but a huge undertaking for one person.

David and I recently had a text conversation during which he shared how he was using ChatGPT to help manage content on the site. He recently did a post called “Shout Outs for Seniors”:

“I collected nominations in a form. Fed the exported form data to ChatGPT, spent about 15 minutes and it created the HTML bookmarks at the top of the page linking to each nominee, the H4 headline tags, etc. Rather than having to do all that by hand.

I’ve got the writing prompts honed in on Claude (rather than ChatGPT) so it writes pretty close to my style. For baseball games I have started just looking at my box score and recording a voice memo recapping what happens. I upload the audio to Dropbox, ChatGPT watches that folder and transcribes it. I feed the transcription to Claude and get a rough draft of my game recap. If I have quotes, I feed it my quotes and tell it to use them verbatim. Make a few tweaks when I’m finished and it’s ready.

The voice cloning really creeped me out when you first mentioned it, but I am paying for an ElevenLabs account. I’m not using my voice yet, because it’s not good enough. But I have tinkered with the API and will probably be adding a “listen to this” audio player to every article at some point in the future. I’ve got it working, but haven’t put it in place and haven’t calculated what it would cost.

Ideally I would be able to append each story with 2 seconds of text in the API: This audio version sponsored by Central Bank.Followed by the article.

The bottom line is AI tools like ChatGPT and Claude are making it possible to accomplish tasks that once required hours David doesn’t have as a one-man operation. And the athletes and their families are the big winners.

RadioGPT

Posted on by

I first read about –and started plays with– this technology in February of 2023. I wondered at the time how long it would be before radio station owners took notice. No time at all, it seams.

Futuri Launches RadioGPT, The World’s First AI-Driven Localized Radio Content

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.

Stations can select from a variety of AI voices for single-, duo-, or trio-hosted shows, or train the AI with their existing personalities’ voices. Programming is available for individual dayparts, or Futuri’s RadioGPT™ can power the entire station. RadioGPT™ is available for all formats in a white-labeled fashion.

RadioGPT™ also generates social posts, blogs, and other content for digital platforms related to the content on the air in real-time. A TopicPulse Instant Video add-on creates AI-driven short videos on hot topics for social use. By adding on Futuri’s POST AI-enabled podcasting system, stations can take broadcast audio and immediately publish it on-demand with POST’s auto-publishing feature.

ChatGPT can now detect emotion

Posted on by

“Among other things, ChatGPT can now detect emotion in both the user’s voice as well as from their facial expression, just like a human can. It also makes unprompted jokes, the way a human would who’s trying to keep a conversation light, and it also lets you interrupt a response — so that you no longer have to confine yourself to the stilted my turn-your turn dynamic of a conversation with a chatbot.”

“If you had your eyes closed, you’d think this is a real lady fawning over a cute puppy, when in fact it’s an AI model that’s learned how to express relevant and appropriate emotion — in addition to making the same observations we would when we meet a cute dog for the first time.”

Two kinds of radio station

Posted on by

Last week the owner of Moberly, Missouri radio stations KWIX and KRES —Alpha Media— laid off all of their on-air staff. (Bob Priddy mourns)

The following is from a post and interview I did in 2007 with Dave Shepherd, the son of the man who put KWIX/KRES on the air.

Fifty years ago, Jerrell Shepherd mastered a form of broadcasting alchemy that turned small town radio lead into gold. It wasn’t much of a secret, however, since he readily shared it with countless radio station owners and managers who made the pilgrimage to Moberly, Missouri, in hopes of bringing some of Shepherd’s sales and programming magic back to their stations.

While most small market broadcasters were content to get “their fair share” of local advertising budgets (the bulk went to the local newspaper), Shepherd’s sales reps were trained to ask for it all and believed in their hearts they deserved it.

Mr. Shepherd’s approach to programming his stations was deceptively simple: report anything and everything that happened in each of the communities covered by his stations’ signals. The KWIX and KRES “Red Rovers” showed up just about every high school football game, junior high choral concert and chamber of commerce ribbon-cutting. And the Shepherd stations put it all on the air. Always with local sponsors. Lots of local sponsors.

The new owner, Alpha Media, owns a lot of radio stations including KBFF Live 95.5 FM in Portland, OR. Last June the station introduced the first AI-powered DJ, “AI Ashley.”

“Alpha Media’s EVP of content Phil Becker assured listeners that Elzinga’s job is safe and she’ll be receiving the same pay, telling TechCrunch that AI Ashley is a tool that will allow DJs to multitask like never before.”

I’d love to know what sort of prompt could result in an AI making a call and interacting with a listener as we heard in the clip above. As the program director of a small town radio station back in the 1970’s I was responsible for hiring and training weekend talent. I might have jumped at the chance to put an AI voice on the air.

Will the KWIX/KRES on-air staff be replaced with AI voices? If so, how will the station’s listeners and advertisers respond?

Perplexity says (using British accent) she has no voice

Posted on by

As good as I’ve found ChatGPT to be, it appears Perplexity (another AI tool) might be even better. One of my favorite features of ChatGPT is ability to “converse” with one of several human-like voices. I posted an example of this a few months back. Others nitpick but I find the voice amazingly human-like. Now that I have a Perplexity account I wondered if it has the same feature.

Does not have a voice feature? Sooo, what are we listening to here. I’m sure there’s a logical explanation for this and I’ll update this post if and when. Perhaps it’s just a new feature (that it doesn’t know about?!)

What I was curious about –before this distraction– was why the ChatGPT voice is so natural sounding and the Perplexity voice is canned computer-speak.