avatarGutbloom

Free AI web copilot to create summaries, insights and extended knowledge, download it at here

4690

Abstract

0/1*jFax3XhfV9mam-k4kI_NSQ.png"><figcaption>A paragraph is summarized into one sentence</figcaption></figure><p id="2e04">Another use of text summarization is to present a user with a auto summarized dialog, with a <i>read more </i>option, which can then expand into the longer un-summarized version.</p><figure id="82f1"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="58c0">Keywords</h2><p id="4f31">Keywords can be extracted from a block of text. You can configure the environment to be conservative and select only keywords from the text. Or a higher <i>temperature </i>can be set to where related words or keywords are generated.</p><figure id="286e"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*FGE3OBMcMyDDcpgKRQb9AQ.png"><figcaption>Key words generated from a Wikipedia paragraph.</figcaption></figure><p id="3a5c">This is very helpful to categorize text and create a search index. In the image above a extract on soccer was taken from Wikipedia. GPT-3 converted this quite large paragraph into six key words or themes.</p><figure id="93e7"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="e562">Parse Unstructured Data</h2><p id="3191">Create tables from long form text by specifying a structure and supplying some examples.</p><figure id="6942"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*X9YWrg_wlpg3hbKCmoq62A.png"><figcaption></figcaption></figure><p id="6984">Here you can see the first entry is directly related to the sentence. The subsequent entries are somehow related and still relevant and applicable.</p><figure id="90e7"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="7585">Classification</h2><p id="72fe">Classify items into categories via example inputs. Companies are named with categories defined. A new company can be mentioned and auto classified.</p><figure id="2265"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*IOkqXlAV2ZCxFpMNaeL2XA.png"><figcaption>With limited training data a new company can be mentioned and auto classified.</figcaption></figure><figure id="9feb"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="281c">Extract Contact Information</h2><p id="0f31">Extract contact information from a block of text. In this case, an address.</p><figure id="8a87"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*5imfXfm2gvrGbA1uRTOkiQ.png"><figcaption>A complete address from the free text message.</figcaption></figure><figure id="ea15"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="95b2">Summarize For A Second Grader</h2><p id="9a5d">This functionality takes a complex and relatively long piece, summarize and simplifies it into a sentence or two.</p><figure id="0da8"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*ReUEmdf0XZ3r54rcYqibfA.png"><figcaption>A large and complex piece of text is summarized and simplified.</figcaption></figure><figure id="d85e"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h1 id="e7ec">Conclusion</h1><p id="3c79">There are definitely good implementation opportunities for the Conversational AI aspect of GPT-3.</p><figure id="6f46"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*5m48Qw7hRcbi7gsL5ZoRAQ.png"><figcaption>Restaurant review is created from a few key words and the restaurant name.</figcaption></figure><p id="2f01">As a support API where text can be processed to assist existing NLU functionality, there is a very real use case.</p><p id="25d7">As mentioned, GPT-3 can be a great help in pre-processing user input as a help for the NLU engine. The challenge is that GPT-3 seems very well positioned to write reviews, compile questions and have a general conversation. This could lead to a proliferation of bots writing reviews, online adds and general copywriting tasks.</p><figure id="b3c1"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*26A6H7nHUvLW10CkKSlaCQ.png"><figcaption>An apple pie review based on four generic words.</figcaption></figure><p id="4fae">This automation does not need to be malicious in principle. Open AI is seemingly making every effort to ensure the responsible use of the API’s.</p><p id="99a2">The fact the extensive

Options

training is not required, and a few key words or phrases can <i>point </i>the API in the right direction, is astounding.</p><p id="a28c">There are however opensource alternatives for most of the functionality available.</p><figure id="04a8"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="3cbc">Positives</h2><ul><li>GPT-3 has quite a bit of functionality which can serve to augment a current chatbot.</li><li>Dialog can be diversified with the NLG capability.</li><li>General chit-chat can easily be created.</li><li>Copywriting is made easy for slogans, headlines, reviews etc.</li><li>Text transformation</li><li>Text generation</li><li>Creating a general purpose bot to chat to.</li><li>With their underlying processing power and data, creating flexible Machine Learning stories should be a good fit.</li></ul><figure id="a4fc"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><h2 id="b03c">Not-so Positives</h2><ul><li>The API is cloud hosted</li><li>Cost</li><li>Social media bot content generation</li><li>Not a framework for sustainable chatbot scaling; <i>yet</i>.</li><li>Possible over and under steering with training data.</li></ul><figure id="bfc0"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure><div id="9145" class="link-block"> <a href="https://cobusgreyling.me/thank-you-for-subscribing/"> <div> <div> <h2>Subscribe to my newsletter.</h2> <div><h3>NLP/NLU, Chatbots, Voice, Conversational UI/UX, CX Designer, Developer, Ubiquitous User Interfaces, Ambient…</h3></div> <div><p>cobusgreyling.me</p></div> </div> <div> <div style="background-image: url(https://miro.readmedium.com/v2/resize:fit:320/0*sxVx6IgC_6AZXIX-)"></div> </div> </div> </a> </div><div id="1141" class="link-block"> <a href="https://cobusgreyling.medium.com"> <div> <div> <h2>Cobus Greyling - Medium</h2> <div><h3>Read writing from Cobus Greyling on Medium. NLP/NLU, Chatbots, Voice, Conversational UI/UX, CX Designer, Developer…</h3></div> <div><p>cobusgreyling.medium.com</p></div> </div> <div> <div style="background-image: url(https://miro.readmedium.com/v2/resize:fit:320/0*e6KH9V073Egac4ua)"></div> </div> </div> </a> </div><div id="2e0e" class="link-block"> <a href="https://cobusgreyling.medium.com/gpt-3-conversational-ai-chatbots-3fb1cfb99942"> <div> <div> <h2>GPT-3: Conversational AI & Chatbots</h2> <div><h3>What Will The Impact Be On Chatbot Design & Development</h3></div> <div><p>cobusgreyling.medium.com</p></div> </div> <div> <div style="background-image: url(https://miro.readmedium.com/v2/resize:fit:320/1*VKDfysUcdNJtCEQwmxSE-w.jpeg)"></div> </div> </div> </a> </div><div id="59bc" class="link-block"> <a href="https://openai.com/"> <div> <div> <h2>OpenAI</h2> <div><h3>OpenAI is an AI research and deployment company. Our mission is to ensure that artificial general intelligence benefits…</h3></div> <div><p>openai.com</p></div> </div> <div> <div style="background-image: url(https://miro.readmedium.com/v2/resize:fit:320/0*SzqTQqBlnX8UOa8A)"></div> </div> </div> </a> </div><div id="47a1" class="link-block"> <a href="https://openai.com/blog/openai-api/"> <div> <div> <h2>OpenAI API</h2> <div><h3>We're releasing an API for accessing new AI models developed by OpenAI. Unlike most AI systems which are designed for…</h3></div> <div><p>openai.com</p></div> </div> <div> <div style="background-image: url(https://miro.readmedium.com/v2/resize:fit:320/0*iD7cRdNC9EDQkZQ2)"></div> </div> </div> </a> </div><figure id="f7c7"><img src="https://cdn-images-1.readmedium.com/v2/resize:fit:800/1*mlG74R9ZBY7NQcvQ2C8dMw.png"><figcaption></figcaption></figure></article></body>

The Clerk

I have to admit that we haven’t gotten much done since securing underwriting by the Russians for the season. I discovered that you can use bitrubles to buy extra lives on the missile command machine in the arcade, so I’ve been going there before lunch to “think about my personal essay” and play games.

So far, the only thing I’ve decided on for my personal essay is the title. I’m going to call it: “Me, me, me.”

You might wonder why I don’t go to the arcade after lunch. Wouldn’t sitting on an arcade stool aid digestion? That’s true, but it is hard to eat an orange push-up while playing video games, and after lunch I like to stretch out on the couch in the office and take my “constitutional” siesta.

Today, the hiss of the electric samovar woke me up. Pee Wee was sitting on his milk crate reading the newspaper.

“Count Pee Wee,” I said, “Why don’t we have our clerk make us some tea? Wouldn’t that be nice?”

“If you are going to give me a Russian Title,” Pee Wee said, “I want to be an earl.”

“The Russians don’t have earls. A count is the equivalent rank,” I said, “Why do you want to be an earl?”

“Because of Earl Monroe.”

“You know,” I said, “You can stop being a Knicks fan anytime you want. You can always switch to the Celtics. Even Spike Lee is watching Wimbledon right now. He was at the Venus Williams match this morning.”

Pee Wee looked back down at his newspaper without replying, meaning he was done. That didn’t stop me from trying to talk to him.

“Tell our clerk to go down to N_____ prospect and buy you a new set of boots. Not the cheap ones, but proper ones that come up to the knee. Before he goes, have him to make us some tea. The samovar is steaming and the afternoon might have a chill in it.”

“We don’t have a clerk,” Pee Wee said without looking up.

“Why not? We have plenty of money. The bitrubles are rolling in, aren’t they? We certainly can afford a clerk.”

Pee Wee didn’t reply. I decided I needed someone else to bother. I went upstairs to the accounting department only to find it closed. As I have explained before, the people in the accounting department are all on the same cycling team. I don’t know why an operation our size needs ten accountants, but I suspect it’s because you need nine riders plus the person who drives the little car filled with snacks to make a team. The Tour de France is on right now. The accountants aren’t like me. Watching a sport inspires them to do the sport, which I think betrays a lack of imagination.

I decided to go to the art department. The only person there was Radio Free Spruce.

“It’s almost done,” he said.

“I’m not here about a deadline, Spruce. I was wondering if you knew why we don’t have a clerk yet.”

“When will you need the answer by?”

The minute I began talking, he said, “That’s too soon. We’ll need at least three days.”

“Spruce,” I said, “This isn’t about anything the art department is doing. I just wanted to know if you knew why we haven’t hired any clerks yet.”

“Sounds like editorial fucked this one up,” he said. “But of course they’re going to blame the art department. We don’t even have clerks. What’s a clerk, anyway?” Then he looked at his watch and said, “Wow, it’s almost four. Do you want to get high?”

It was 2:15. I decided to go back downstairs to the office. When I got there, Pee Wee was asleep on the couch and the samovar was still hissing.

I decided to take out an ad in Mushamaguntic’s Auto Trader newspaper. My thinking was, if someone was so hard up that they were looking for a car in Auto Trader, they might be desperate enough to take a job with us.

Here is the ad I placed:

The Mill is Hiring We Are Looking for 1 Clerk The ideal candidate will be able to handle a slow-paced environment without become bored, sarcastic, or bitter. Must be able to pick up the phone. Knowledge of assembly language, canon law, and quantum mechanics a plus. Ability to generate high quality “content” at a moment’s notice is a must. No Mollusks need apply. Salary Commensurate with cribbage skill.

The first, and only, person to apply was a woman named Claire who used to work at the Mushamaguntic Normal School. This is how the interview went:

Gutbloom: Didn’t you apply here once before?

Claire: Yes, I applied to be an intern, but you gave that job to Ernio Hernandez.

Gutbloom: Well, I had to give him the job.

Claire: Why?

Gutbloom: I thought he might be related to Hector Hernandez, one of the founders of H&H Bagels. If he was, he would know the secret of rolling bagels. Do you know how hard it is to get a good bagel up here?

Claire: Why didn’t you just ask him if he was related?

Gutbloom: If he knew the secret, he would lie about it, wouldn’t he?

Claire: So, does Ernio know anything about bagels?

Gutbloom: I’m not sure. I’m still trying to figure that out.

Claire: You understand that Hernandez is the 29th most popular name in the US, don’t you? It’s right behind “Young” and above “King”.

Gutbloom: I only know two. Ernio and the guy from the bagel place.

Claire: Maybe we should start my interview.

Gutbloom: Well, there’s a big problem.

Claire: What’s that?

Gutbloom: I don’t think I can hire you.

Claire: Why not?

Gutbloom: Well…. you see… I mean it’s a blog, and I end up having to describe everyone physically, and it’s clear that you are… I mean… you’re not…

Claire: Are you refering to my size?

Gutbloom: Exactly!

Claire: What’s the problem?

Gutbloom: It’s like a trap of tropes. If I make you funny, then people will be like “oh, the funny fat girl”, and if I make you greedy, lonely, oversexed, undersexed, neat, or messy, I’ll probably get 10,000 letters from “your fat friends” telling me what an ass I am. It’s a metafictional nightmare.

Claire: Why don’t you just describe me honestly without focusing on my appearance? Look, we are well into my interview and the only thing you have described is that I’m heavy. You haven’t even mentioned that I’m wearing a tee shirt that says, “My dick is made of cheddar cheese.”

Gutbloom: I was going to get to that. [editor’s note: Claire is wearing a high quality black tee-shirt with “My dick is made of cheddar cheese” written in yellow block letters on the front. Her hair is brown and shoulder length. She is not wearing glasses. Who wants a clerk that doesn’t wear glasses?”]

Claire: I’m wearing contacts.

Gutbloom: Oh. I’m going to continue the interview, but I still think the weight thing is a problem.

Claire: It’s not a fucking problem. Didn’t you have a character named “Fat Joe” last year?

Gutbloom: Yes, but he was clearly an amalgam of my Irish-American self-hatred. Even then I got in trouble for making him smell good. All three people who read that post were like, “Fat Joe smells good? Bullshit.”

Claire: Just give me the whiteboard problem, will you?

Gutbloom: Wait. One more thing. I’m not going to ask you about your ethnicity because the HR people tell me if I ask any questions about race I’m fucked, but if you are black there is no way I can give you the job.

Claire: I think I either just got the job or a winning lawsuit. Why wouldn’t you hire me if I were black?

Gutbloom: How am I going to write a black character? I guess I could try to make you a black person from Martha’s Vineyard or that went to St. Pauls, but it would be the blogging equivalent of William Styron writing The Confessions of Nat Turner. Even if I’m successful, I get accused of “appropriating” black culture. Like, say I have you deconstruct Beyonce’s Lemonade

Claire: I don’t understand Lemonade. I think I might have some inkling about why it is so powerful, but I wouldn’t want to share my thoughts out loud.

Gutbloom: So, I shouldn’t have mentioned Lemonade?

Claire: I can’t imagine why you would bring it up, but I can fix your weird middle-aged, middle-class, white guy authorship problems. I am an alien.

Gutbloom: Oh, no.

Claire: From the planet Xenon.

Gutbloom: Oh, thank goodness! What religion are your… people.

Claire: Xenonics are orthodox followers of the Flying Spaghetti Monster.

Gutbloom: Who is allowed to speak for the Xenonics?

Claire: Anyone who wants to, and anyone who has touched one of us is considered a member of the tribe and their mockery will thereafter be understood as “self-deprecating.”

Gutbloom: Welcome to the Mill! [shakes hands]

Claire: That was easy. I didn’t even break a sweat.

Gutbloom: Thanks goodness. Please don’t sweat.

Claire: Why not?

Gutbloom: How am I going to describe the fat girl sweating without being an asshole?

Claire: Why would you have to mention it?

Gutbloom: The sweat glisening on the wide expanse of skin that is your young chest, how could I not?

Claire: Look [holds wrist in front of me] I am wearing an Apple watch. Why don’t you talk about that? I’m sure your idiot readers…

Gutbloom: Our idiot readers. You’re on the team now.

Claire: …the idiot readers are more interested in my Apple watch than the sweat on my body.

Gutbloom: You overestimate our readers. For the record, I noticed the Apple watch. I was thankful it wasn’t a fitbit.

Claire: [rolls eyes] Wow. You are such an asshole. Where do I put my stuff and when are we going to negotiate my salary?

Gutbloom: Your salary is based on your cribbage skill.

Claire: I don’t play cribbage. I play bridge.

[Pee Wee grunted audibly. He has been looking for a new bridge partner since Ester “ice bid” Payson retired to Florida.]

Gutbloom: Oh, fuck!

Claire: What’s the problem?

Gutbloom: I don’t understand bridge. How am I going to write about you and Pee Wee playing bridge?

Claire: That’s your problem, isn’t it? Pee Wee, would you like some tea?

Humor
The Mill
Dreck
Funny
Fiction
Recommended from ReadMedium