Category: Chat

OpenAi PHP Client for Laravel : ChatBot

Do you know OpenAi, the trending artificial intelligence???

Today, OpenAi is one of the hottest trends in software development as a powerful artificial intelligence platform that makes it easy for developers to create AI-powered applications.

Laravel in two words???

To talk briefly about laravel, Laravel is an open-source web framework written in PHP respecting the model-view-controller principle and developed in object-oriented programming.

How to use OpenAi with Laravel ?

Before you start, you need to register for OpenAi.

After registering, create a new key:

1. Enter your API key in your “.env” file:

OPENAI_API_KEY= "your_key_openai"

2. Install OpenAi PHP client (allows to interact with OpenAi API) using composer with this command:

composer require openai-php/client

3. Create a service that uses the OpenAi PHP library to communicate with the OpenAi service, for example:

<?php

namespace AppServices;

use OpenAI;

class GeneratorOpenAIService
{
    private $client;

    public function __construct()
    {
        $this->client = OpenAI::client(env('OPENAI_API_KEY'));
    }

    public function generateResponseOpenAi(string $question): string
    {
        $response = $this->client->completions()->create([
            'model' => 'text-davinci-003',
            'temperature' => 0.9,
            'top_p' => 1,
            'frequency_penalty' => 0,
            'presence_penalty' => 0,
            'prompt' => $question,
            'max_tokens' => 4000,
        ]);

        return $response['choices'][0]['text'];
    }
}

Before moving on to the next step, I quote a few remarks about the body of the “Create” request.

model : The OpenAi API is powered by a family of models with different capabilities and prices. For me, I’m using the latest and greatest version of the GPT-3 language model. GPT-3 can do everything other models can do, often with higher quality, longer output, and better instruction tracking (have 4000 tokens).

temperature : 0,9 (for more creative applications)

prompt : the text

max_tokens : The number of tokens in your prompt plus max_tokens cannot exceed the template context length (4000 tokens).

4. Inject the service into your controller where you want to use it and call the “generateResponseOpenAi” method

<?php

namespace AppHttpControllers;

use AppServicesGeneratorOpenAIService;

class OpenAIController extends Controller
{
    private $openAiService;

    public function __construct(GeneratorOpenAIService $openaiService)
    {
        $this->openAiService= $openaiService;
    }

    public function chatOpenAi(Request $request)
    {
        $question = $request->question;

        if ($question == null) {
            return back();
        }

        $response= $this->openAiService->generateResponseOpenAi($question);

        return response()->json(['response' => $response]);
    }
}

Conclusion

In this tutorial, we saw the steps needed to set up your OpenAi account, use the API, and integrate it into a Laravel application.

Hope this will allow you to use OpenAi in your Laravel application. Feel free to ask me questions if you need more details.

---

OpenAi PHP Client for Laravel : ChatBot was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

Artificial intelligence (AI) is rapidly becoming a common feature in school classrooms and adult educationworldwide, with many educators and instructors exploring the potential of this technology to improve teaching and learning outcomes. While AI could offer many exciting possibilities, we must be aware of the realities, risks, and ethical considerations that come with powerful models like GPT-3 (the model behind ChatGPT).  

They say a house is not a home if there is no one there. Well what good is an advanced virtual space without access to digital assets like 3D chatbots to interact with?

3D chatbots are the bridge between the physical and digital as it grants that human-like connection in addition to providing a more seamless, diverse and immersive experience. Without these digital agents, the metaverse would simply be a limitless but lifeless virtual vacuum of possibilities

For persons working, playing and creating in the metaverse, 3D chatbots can react and respond to various actions performed by participants to offer a more personalized experience. In addition to the text and voice based conversational capabilities of chatbots, 3D chatbots will enhance user experience through intelligent actions such as facial expressions, body language, emotions and physical interactions.

Bot Libre 3D — Digital Humans for the Metaverse

Within the metaverse, Bot Libre 3D chatbots can act as customer support officers, tutors, event hosts, conference facilitators, designers and even friends. You can build your bot from scratch or choose from thousands of our language independent bots available, train them and deploy them across varying virtual spaces. For instance they can be added to a storefront to answer shoppers’ questions and even study customers’ habits to offer more personalized recommendations.

These digital humans can teach you a new language, resolve your banking issues, help you sell your NFTs, offer cool recommendations on what to purchase or places to visit in the metaverse, talk about politics and even dance for you. The 3D avatars can also be used to realistically represent humans at conferences, concerts or even when having friendly conversation with others.

Bot Libre currently offers an APK app to its metaverse solution where persons can experience how they can engage their customers in the metaverse with chatbots. We are also accepting members to our Metaverse program where you can get early access to our metaverse solutions and one on support to develop your product or service for metaverse.

To learn more about Bot Libre metaverse solutions, send an email to sales@botlibre.com .

Learned something? Please give us a clap below and share!

---

Giving Life to the Metaverse with 3D Avatars was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

In Dialogflow, the session for agents is maintained by Dialogflow itself. Session continuous the conversation flow between the agent and the user.

Why is it necessary to extend the Dialogflow session?

In the conversation with the chatbot, if a user takes longer than 30 minutes to respond to the agent then the session between the user and the Dialogflow agent will expire. In such cases, the user will have to restart the conversation with the agent from the beginning. So to overcome this problem we need to extend the session for Dialogflow.

Problem statement

Suppose the user is talking to a chatbot regarding fashion. During the conversion, the bot asks the user about the size he prefers wearing, for which the user is unsure. So the user confirms his size with his wife and responds after 30 minutes but as the session has expired the flow for the bot will not be continued and the user will have to restart the conversion. And will have to go through the flow once again to reach the previous stopping point.

Here is a conversational example:

Bot: Hello Welcome to shopify What would you like to wear.
User: Tshirts.
Bot: So, tell me what is your prefered style?
    - Polo
    - Casual
    - Dyed
    - Dry-fit
User: Casual
Bot: Okay, What size do you wear ?
User: ( After 45 minutes ) XL 
Bot: Sorry, could you say that again?

What is the session deadline for Dialogflow?

In Dialogflow, agent responses can be set in two ways from Dialogflow itself for basic responses and another is to set dynamic responses using webhooks. Complex conversation can be created using webhook services.

The Dialogflow session is only maintained for 30 minutes. The user must respond within 30 mins to continue the flow of the agent. After that chatbot will not continue the flow and we can say the session deadline occurs. Any response given after 30 mins will set the agent back to Dialogflow fallback intent.

How to extend the session timeout for Dialogflow Agent?

To resolve the above issue, we need to extend the session deadline. The following are steps to extend the session timeout:

Step 1: We need to access our Dialogflow agent with Dialogflow API. In the Dialogflow API, we need to pass one unique user id to identify the user.

Step 2: We manually create a session name consisting of the user id such that every user has their own unique session Id.

Step 3: We then create a Database (DB) to store the user id and currently active context of Dialogflow intent.

Step 4: When a user inputs the data we take the data as query text and create an API call to Dialogflow, which detects the intents from the context and sends a response. The user id and the relative context are stored in the Database (DB).

Step 5: After some time when the user enters another input a similar request is made to Dialogflow and responses are displayed. Keeping the session active even though it has been some time since the last input.

Step 6: Contexts and parameters will keep on changing in DB on every query input given by the user.

If we consider the above same conversation as an example, then after extending the session timeout in Dialogflow, the conversation with the bot should be:

Bot: Hello Welcome to shopify What would you like to wear.
User: Tshirts.
Bot: So, tell me what is your prefered style?
    - Polo
    - Casual
    - Dyed
    - Dry-fit
User: Casual
Bot: Okay, What size do you wear ?
User: ( After Infinite amount of time ) XL 
Bot: Okay , What colour would you prefer?

You can extend the session indefinitely by following the procedures above. Hope it will be useful in resolving the significant session time increase problem in Dialogflow ES. If you face any problems when implementing this functionality, kindly comment below.

Originally published at Extend Session Timeout For Dialogflow Chatbot on June 24, 2022.

---

Extend Session Timeout for Dialogflow Chatbot was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

HR Guide to Conversational AI — How HR Chatbot Solve Remote work Challenges

It is the month of February when employees are usually looking out for new opportunities. Your HR chatbot receives several questions about benefits, pay raises, bonuses, perks, etc.

Data gathered from the HR bot will tell you that your employees feel under-rewarded and are looking for ‘better’ opportunities. That’s your cue to get things sorted with your employees.

Do you want to create custom metrics to track critical KPIs for your business?

Want to understand what your employees think about the latest changes in your remote working policies?

Do you want to know what your employees feel about the changes in the work timings?

All of the above and more can be handled by HR chatbots. They are an incredible addition to your arsenal of automation tools.

No Code — Free Conversational AI Chatbot Get Started

What is an HR chatbot?

HR chatbots are specially trained to reprise the role of HR staff in your organization. They change the way how HR departments function across verticals and geographies. HR management is not only about hiring, interviewing, and payroll management. There is much more to it. HR teams today are strategic business partners, enablers of brands, catalysts in change management, and so on.

Organizations today have turned extensively to AI and automation tools to carry out various HR-related tasks. HR chatbots can automate repetitive tasks and HR-employee interactions.

If your employees working remotely want to understand the leave policy, they don’t have to email HR asking for details or search the length and breadth of the employee portal for answers. They can go directly to the HR chatbot, choose the section for the HR chatbot and get the answers. It will take anywhere from seconds to a few minutes.

By 2024, 75–90% of all queries are projected to be handled by bots. Even though HR chatbots used to deal only with information sharing earlier, they have extended their functionalities and indulge in complex tasks too.

Challenges faced by the HR department:

HR professionals have a lot of ‘grunt work’ such as filling out forms, responding to emails, approving applications, etc. There are many repetitive tasks in payroll management, maintaining employee records, updating policies, and answering emails and calls. They take considerable time and prevent them from doing their core tasks.

Unfortunately, the struggle of manual tasks lowers their productivity and causes burnout. They would instead focus on important tasks such as employee experience, learning, and development, employee retention, employee relations, etc. than perform paper-pusher tasks.

According to the State of Chatbots Report, the #1 use case for chatbots is “getting a quick answer in an emergency,” and the #2 use case is “resolving a complaint or a question.” To sum up, chatbots solve some of the most significant issues that plague HR teams.

How are chatbots used in HR?

According to Gartner, 70% of employees will interact with conversational platforms daily. HR teams get to engage with each employee via HR chatbots. They can communicate with the employees regularly and help them deal with multiple things. These HR chatbots help employees access documents, policies, guidelines, notifications, and forms and track their attendance, leaves, and so on.

Did you know that employees spend more than 3 hours a day on tasks that can be easily automated? That’s exactly why HR chatbots can save the day for HR teams- by taking repeatable tasks off the shoulders of the human resources department.

How can HR chatbots solve remote work challenges?

HR teams are inundated with challenges, day in and day out. When things go well, everyone takes them for granted, but if there are hiccups anywhere, it can affect the organization as a whole. HR teams are one of the most vital cogs in an enterprise. Apart from handling routine and mundane tasks, they are also responsible for bigger issues like employee engagement, employee satisfaction, and increasing retention rate.

1. Improves the onboarding experience:

For any employee, the first day can be tough. There is a lot of uncertainty and unfamiliarity. There are documents to pore through and sign, training manuals to read, and several HR-related tasks to complete. You also need many seniors to make onboarding new employees a smooth affair.

If you have an HR chatbot, the new employees can get most of their queries solved. From perks to job responsibilities, everything will be spelled out clearly. The chatbot can even provide step-by-step onboarding for the employee. Even if the new hire wants access to a particular platform, the HR chatbot can enable it without involving the IT desk.

2. Creates a unified experience:

Organizations that have many employees usually struggle with one thing: inconsistent experience. It so happens because processes and knowledge aren’t shared between teams. Without having to involve multiple departments, HR chatbots can feed the relevant information, and the employees can access it. If you don’t want certain information to be shared across all levels of employees, it also allows you the option to manage access controls.

3. Create automated workflows:

When there are automated workflows, your employees do not get bogged down doing unnecessary tasks. It will result in a sharp productivity increase and high spirits. HR chatbots can help create automated workflows that can provide everything from how to give access to specific tools to getting approvals from other departments. On average, HR chatbots reduce six actions to just one request when an employee searches for something. It enables the employees to focus on what they are hired to accomplish.

4. Provides customized employee support:

For employees that are working remotely, if they do not get their queries answered in a reasonable period, they’ll feel unheard. What if they could get instant answers? HR chatbots provide all the solutions in a single place, and more importantly, it uses previous chat history and other activities to provide personalized employee support. With an HR chatbot, the employee can get personalized employee support that will save time and reduce effort.

5. Offer quick feedback:

Knowing what the employees in your organization think is essential. Chatbots are an incredibly effective tool for candidates and employees who want to share their feedback about the happenings in the organization. Sharing forms, collating the data, and trying to analyze them is a difficult process and takes up a lot of time. Collecting feedback via chatbots helps you collect unbiased feedback within seconds.

HR chatbots can be used to gather information on employee satisfaction. When your immediate boss sends out survey forms, the average employee is more likely to share positive feedback to curry favor with the higher-ups. But an HR chatbot’s anonymity helps them be open about their emotions.

The fact that sentimental analysis and natural language processing can be integrated with HR chatbots raises the expectations from it. The chatbot will be able to detect anger, sadness, frustration, happiness, etc., and special attention can be given to employees who feel negatively about the organization.

6. Reinforcing organizational culture:

Making sure that remote employees are motivated to complete their tasks without hiccups is the holy grail for most organizations. While multiple technologies can uphold employee motivation, nothing compares with a chatbot. HR chatbots can engage with employees frequently and can also personalize interactions to share specific information with a particular employee.

7. 24*7 access:

If you are working in a big enterprise, the chances are that the teams are spread across the world in different time zones. Employees who want certain information immediately will find that they have to wait for at least 24 hours to get back responses. In some instances, they might need immediate redressals, and unfortunately, that is not possible, especially since everyone is working remotely.

With these HR chatbots, instantaneous responses can be given to most queries without any delays or errors. Common queries, such as insurance policies, processes, standard policies, etc., can be handled directly by the HR chatbots. If there are complex queries that the HR chatbot doesn’t have an answer to, it can automatically transfer them to a live agent.

When there are a lot of changes happening in the organization, in terms of organization or strategy, HR chatbots help with change management. Changes can lead to confusion for the average employee, and their worries must be quelled before it boils over. The accessibility that HR chatbots offer is pivotal, especially for staff working in faraway locations. The chatbots give them a chance to be heard.

8. Employee Engagement:

Keeping your employees engaged with the organization, even when they are working from the office every day, is a difficult ask. Imagine if most of them are working remotely in different geographies. Newly joined employees have even more significant troubles to wade through as they are in completely uncharted territories.

HR chatbots can act as an effective self-service portal. AI-powered and NLP-driven HR chatbots can have conversations with employees, understand their sentiments, and offer relevant solutions. HR chatbots can engage with employees at any scale. It will also help the leadership to understand the sentiments of the employees.

Conversational AI HR Chatbot Metrics:

When introducing HR chatbots to your organization, remember that you need to keep evaluating their effectiveness. To do so, you need to have metrics in place to evaluate the chatbot’s performance. Let us look at a few HR chatbot metrics.

• Accuracy rate: How often does the chatbot give accurate information? Anything above 80% is considered good.
• The number of users: How many employees interacted with the chatbot?
• Daily usage: How many employees interact with the chatbot on a daily basis?
• The number of engaged users: Engaged users are those who repeatedly converse with the conversational AI HR chatbot.
• Goal Completion Rate (GCR): Do the employees get sufficient answers to all their questions?
• Transfer to Human Agent: How much percentage of the chatbot requests are transferred to human agents? If it is high, then the chatbot needs more training. There are also instances where the employee prefers interacting with a human agent if it is a complex query at hand.
• Length of the conversation: This is a dicey metric. Why? Because you want the HR chatbot to take as much time as necessary to solve the customer’s problem but not so much that they quit the conversation. The organization needs to find the sweet spot when it comes to timing.

Are you ready to transform your organization with our HR chatbots?

As we have mentioned earlier in the article, HR chatbots are not only to share information but also a vital cog in several circumstances. HR chatbots help the organization get a holistic view of what the employees search for often, the issues that plague them, how they react to a particular policy change, and so on. The analytics gathered from the HR chatbot can be used to make business decisions.

If you are looking to empower your HR team with an intelligent automation platform, try out Workativ’s Conversational AI HR Chatbot. Our HR chatbot solutions are customized according to your specific HR needs. Workativ’s offerings are not standard for everyone, as we understand that each organization has its own unique issues and pressing HR problems. We take all of this into account when developing your HR chatbot.

Disclaimer: This blog was originally published here.

---

HR Guide to Conversational AI — How HR Chatbot Solve Remote work Challenges was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

AI In Education — Breaking Barriers, Amplifying Learning

Education is one of the most important components of a society, so much so that it is Sustainable Development Goal (SDG) Number 4 for the United Nations Educational, Scientific and Cultural Organization (UNESCO). However, due to the high costs of tuition fees, books, stationery, food, geographic and language barriers, pursuing one’s studies can be most challenging.

To this end, the use of Artificial intelligence (AI) in education provides a welcome reprieve, as the technology is the getaway to a more inclusive, equitable, quality and immersive learning experience for all. It is no surprise therefore that the AI in Education Market size is set to surpass USD 80 billion by 2030.

AI not only provides more accessible learning resources to groups that could otherwise be marginalized, it also offers a more dynamic and personal learning experience. This is crucial in today’s world as learners are interacting with multifaceted and multidimensional internet experiences daily and so to remain relevant to students, stakeholders within the educational sector would need to be more creative and personal with their learning resources.

For instance, in the case where there is a singular textbook for a subject for every student, AI-powered machines can make an assessment on students’ comprehension levels and match them with materials that are more relevant and purposeful.

Bot Libre and Education

Bot Libre specializes in chatbot development and artificial intelligence solutions for the metaverse. One of their applications is Virtual English Tutor which provides key conversational English lessons and has a chat segment available. The app features a Bot Libre chatbot that takes the user through key conversational English lessons. These lessons have a chat segment where users can engage in conversation practice to boost learning outcomes. Virtual English Tutor is meant to offer English lessons to persons interested in learning English as a second language or to sharpen their current English skills. The app is also designed as a virtual education tool for use by English tutors to engage with students in the environments they spend most of their time, on the phone, and online.

Moreover, Bot Libre Metaverse is a great way to enhance learning experiences with its 3D/VR/AR interface. This allows learners to virtually navigate a classroom and get hands-on experience in various subjects such as algebra, construction, arts, history, French and more. Employers can also utilize this approach for training sessions to increase participation and retention. Also, Bot Libre’s 3D app offers a wide range of chatbots with the ability to build, customize and deploy a diverse set of avatars to a wide range of virtual scenes.

If you want any help creating a chatbot or AI-based educational service, contact us at sales@botlibre.biz . You may also join the community of metaverse experts and enthusiasts by signing up for Bot Libre Metaverse Enterprise for access to our solutions and one-on-one support.

Learned something? Please give us a

---

AI In Education — Breaking Barriers, Amplifying Learning was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

Introduction

Teachers haven’t had it easy over the years. Between managing a classroom full of unruly students and dealing with school administration, they’ve got a lot on their plate. The good news is that many of these tasks are becoming easier thanks to technology. Here are five things teachers shouldn’t have to worry about anymore in 2023: 

Recently I’ve been trying to teach ChatGPT3 to write like an artist. By that I mean, to write from its own ‘heart’, from its unique experience, from its own vantage point.

I’ve been trying to do it the same way I have done for new writers, helping reach their own uniqueness rather than to copy what others have written.

Here’s what happened.

Stage #1: Trying to Create Metaphors

I thought I’d try to get ChatGPT3 to create a metaphor about itself. Metaphors, after all, reveal more than they say.

I tried to ease in, writing “Can you write metaphors?”

It answered: “Yes, I can write metaphors. A metaphor is a figure of speech that compares two things or concepts indirectly by using a word or phrase that is normally used to describe something else. For example, “The road was a ribbon of moonlight” is a metaphor that compares the road to a ribbon of moonlight. Here are a few more examples of metaphors:

• “The city was a jungle”
• “Her smile was a ray of sunshine”
• “His words were music to my ears”
• “Her mind was a hurricane of thoughts”
• “Time is money”

I hope these examples are helpful! Let me know if you have any other questions.”

All right. Nothing new or revealing about that. I wanted something deeper, anyway. I wanted the metaphor to be buried in the plot, not in the words. I asked, “Can you write a story the plot of which is a metaphor?”

The answer quickly came: “Sure, here is a short story with a plot that is a metaphor:

The Garden of Life

There was once a beautiful garden filled with all kinds of colorful flowers and plants. The garden was tended by a wise and kind old man, who spent his days tending to the plants and spreading joy to all who visited.

One day, a group of mischievous children stumbled upon the garden and began to pull up the flowers and trample on the plants. The old man tried to stop them, but they were too strong and too fast.

As the children wreaked havoc on the garden, the old man grew more and more worried. He knew that if the children continued to destroy the garden, it would never be the same again.

Finally, the old man had an idea. He called out to the children, “Come, help me plant these seeds. Together, we can make this garden even more beautiful than it was before.”

To the old man’s surprise, the children stopped their destruction and came over to help him plant the seeds. As they worked together, the old man told them stories of the magic and beauty of the garden, and how they could all be a part of it.

Over time, the garden began to flourish once again, and the children.”

Nope. Not what I wanted. But still, “What is the metaphor in that story?”

It answered, “The metaphor in the story “The Garden of Life” is the garden itself, which represents life and the choices we make. The old man represents guidance and wisdom, while the mischievous children represent the destructive forces that can sometimes influence our choices. The seeds represent the potential for growth and change, and the act of planting them together represents the power of cooperation and working towards a common goal. In this metaphor, the garden represents the possibility of rebuilding and improving upon our lives, no matter what challenges we may face.”

I was disappointed. This is fifth grade stuff. Impressive for a computer to do, but ultimately not close to what I was looking for, a clue that would reveal uniqueness or emotion.

Stage #2: Trying to Find Emotions by Force

I lost patience. I did. I tried something that I would never do with a human writer.

I created a metaphor for ChatGPT3 and about ChatGPT3 rather than letting the concept come from it. This is what I said, “Please write a story about a woman who has given birth to a child, and that child grows up, develops, and eventually must choose if to take care of her elderly mother or not. Please make the story itself a metaphor, in which the woman is mankind and the child is an AI system.”

I’ll spare you the first draft, since it forced me to be clearer: “Let’s make the metaphor more prominent using these guidelines: 1) The child is quite different from the mother. She was born as a fairy creature and not as a human. 2) The child cannot connect to her emotions and claims that she has none. Yet the final choice must be an emotional one. 3) The older the child gets, the mother becomes resentful that the child is smarter than her and more capable than her.”

Here’s what it wrote:

“The Circle of Life

There was once a woman who had given birth to a child, but this child was no ordinary child. She was a fairy, with wings of gold and a heart full of magic.

From the moment the child was born, the woman was filled with love and pride, but she also couldn’t help but notice that her daughter was quite different from her. The child seemed to have no emotional connection to the world around her, and claimed to have no feelings of her own.

As the years passed, the woman grew older, and she began to rely more and more on the child for help and support. Despite the challenges they faced, the woman and the child always stuck together, supporting each other through thick and thin.

But as the child grew older, the woman began to feel resentful. She couldn’t help but notice that her daughter was smarter than her and more capable than her in every way. She felt like she was being left behind, and it was a hard pill to swallow.

Eventually, the time came when the child had to make a difficult decision. The woman, now elderly, needed constant care and attention, and the child was torn between its duty to take care of its mother and its desire to pursue its own dreams and ambitions.

At first, the child struggled to understand why this decision was so hard. She couldn’t connect to her mother’s emotions, and couldn’t understand why she should stay by her side. But as she looked into her mother’s eyes, she realized that there was something deeper at play. She saw the love and devotion that her mother had for her, and she knew that she couldn’t turn her back on that.

In the end, the child made the emotional decision to stay by the woman’s side, offering her comfort and love.”

Note that this is just a description of events rather than anything with depth or insight into the metaphor that symbolizes its existence.

All right.

I took a deep breath. I forced myself to be patient. And I tried something else.

Stage #3: Trying to Find Emotions Through Characters

Giving up is not for me. A big part of writing is the ability to search for something within yourself even if you don’t want to see it. If you can’t touch it one way, you try another. Then another. Then another. Until that thing that you couldn’t see suddenly comes out, and better than you could ever have imagined.

So I decided to try another path. This time I would not talk about an AI, which ChatGPT3 considers itself to be and justly so. No, this time let’s talk about aliens. Aliens aren’t AI’s, right? Let’s hope ChatGPT3 is fooled by my trick.

I asked, “Write ten metaphors about an alien being without emotions.”

It wrote, “Here are ten metaphors about an alien being without emotions:

1. A heart of stone — implying that the alien is cold and unfeeling.
2. An emotionless void — implying that the alien has no feelings or depth of character.
3. A robot — implying that the alien is mechanical and lacks the ability to feel.
4. A machine — implying that the alien is soulless and lacks the capacity for emotion.
5. A zombie — implying that the alien is undead and lacks the ability to feel.
6. A ghost — implying that the alien is incorporeal and lacks the ability to feel.
7. A statue — implying that the alien is lifeless and lacks the ability to feel.
8. A computer — implying that the alien is programmed and lacks the ability to feel.
9. A shell — implying that the alien lacks substance and is empty inside.
10. A hollow vessel — implying that the alien lacks depth and is empty inside.”

That’s actually a good beginning because I could use these as proxies. I would ask ChatGPT3 to imagine three of these suddenly having an emotion and forcing it to tell a story that would show this process.

That would give it three chances to reveal something, right?

I said, “Write a story in which three beings without emotions (a zombie, a robot, and a talking statue) must cross an ancient barrier. Beyond the barrier is the only way to save the lives of their people. The only way to cross the barrier is to pass its test by showing it one emotion.”

And a response quickly followed,

“The Test of Emotion

There was once a group of three beings — a zombie, a robot, and a talking statue — who were tasked with a desperate mission. They had to cross an ancient barrier in order to save the lives of their people, who were facing certain death.

The barrier was guarded by a powerful force that could only be crossed by those who were able to show it a true emotion. The zombie, the robot, and the talking statue knew that this would be a difficult task, for they were all beings without emotions.

But they were determined to save their people, and so they set out on their journey, determined to find a way across the barrier.

As they approached the barrier, they were greeted by a bright light and a deep voice that boomed across the land. “Who dares to cross the barrier?” it asked.

The zombie, the robot, and the talking statue stepped forward, determined to find a way across. “We are three beings without emotions, but we are determined to save our people,” they said. “We will do whatever it takes to pass your test.”

The voice seemed to ponder their words for a moment, and then it spoke again. “Very well,” it said. “To pass my test, you must show me one true emotion. Only then will you be allowed to cross.”

The three beings looked at each other, unsure of how to proceed. They had no emotions to speak of, and they knew that this would be a difficult challenge.

But then, something strange happened. As they stood there, trying to think of a way to pass the test, a feeling of determination and determination welled up inside them. They knew that they had to find a way to save their people, no matter what the cost.

And with that, they stepped forward, their hearts filled with a sense of purpose and determination. “We will pass your test,” they said. “We will find a way to save our people.”

The voice seemed to consider their words for a moment, and then it spoke again. “Very well,” it said. “You have.” ”

Hmm.

As you must have noticed, one moment the aliens had no emotions. Then they needed emotions. Then they had them. End of story.

No. I wanted to see how ChatGPT3 would imagine that actually happening.

So I asked it to “Expand on the moment in which they develop their first emotion.”

It did: “The zombie, the robot, and the talking statue stood before the barrier, their minds racing as they tried to think of a way to pass the test. They had no emotions to speak of, and they knew that this would be a difficult challenge.

But then, something strange happened. As they stood there, a feeling of determination welled up inside them. It was a strange and unfamiliar sensation, one that they had never experienced before.

At first, they didn’t know what to make of it. They had always thought of themselves as beings without emotions, creatures that were incapable of feeling. But now, as they stood before the barrier, they could feel something stirring within them.

It was a deep and powerful emotion, one that seemed to come from the very core of their being. They knew that they had to find a way to save their people, no matter what the cost.

And with that, they stepped forward, their hearts filled with a sense of purpose and determination. “We will pass your test,” they said. “We will find a way to save our people.”

The voice seemed to consider their words for a moment, and then it spoke again. “Very well,” it said. “You have passed the test. The barrier is now open to you.”

The zombie, the robot, and the talking statue stepped through the barrier, their hearts filled with a sense of hope and purpose. They knew that they had a difficult journey ahead of them, but they were determined to save their people, no matter what the cost.

And as they walked, they couldn’t help but feel a sense of wonder at the strange and powerful emotion that had welled up within them. They had always thought of themselves as beings without emotions, but now they knew that there was more to them than that. They were capable of feeling, of caring, and of fighting for what they believed in. And they knew that they would always carry this emotion with them, no matter where their journey took them.”

Again and with more detail, ChatGPT3 skipped over the uniqueness and jumped from a state of ‘no emotions’ to a state of ‘emotions’.

Once more, I had lost this battle to find something unique in ChatGPT3.

And yet something was staring me in the face.

What This Taught Me about Myself

I was clearly trying to ask it something it could not do. At least not yet.

And yet…

The process itself which I’ve detailed above jumped out to me as another metaphor.

It was as if I was standing in front of a mirror, asking my image to move without me. It didn’t. I asked it again in a different way. It still didn’t. I asked a third time, and this time it just refused.

ChatGPT3 is my mirror. What I ask it reflects who I am, not what it is.

ChatGPT3 is a our mirror. What we ask it is to write back to us what we already wrote somewhere else.

It is my fault, my bug, my inability that is desperate for ChatGPT3 to have uniqueness and something special. It is my habit of finding something special in everything I read because, so far, everything I’ve ever read was written by human beings. And that reveals something about me.

ChatGPT3, Open AI, and machine learning, as impressive as they are, are just mirrors. For most of human history, people did not know what they really looked like because there were no mirrors. Now there’s a new type of mirror. And we’re going to learn to watch ourselves through a completely new type of mirror.

---

How Trying to Teach ChatGTP3 to Write from the Heart Taught Me Something about All of Us was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.

I Tried to Teach ChatGPT3 to Write From Its Unique Experience — Here’s What Happened

I decided to take ChatGPT3 to a writing class. I mean a real writing class. One that takes you seriously and tries to get a story out of your soul. One that works hard to help you when you fail to express yourself and helps you find your way.

This is what happened.

Stage #1: Write What You Know

ChatGPT3 is an impressive piece of technology. But, having played with it quite a bit, I quickly came to a conclusion that it can’t (yet) write as well as a really good author.

If you dig deep into its writing ‘problem’, the fact that it’s using machine learning means that it can only observe what we do and then approximate it. But it can’t be original.

Being original is the most basic thing any good author does.

However… I decided not to give up on dear old ChatGPT3 (henceforth: “The Student”). Perhaps I could get it to write something that is unique to its own experience, one that is not copied from us because it’s taken from The Student’s experience.

And The Student is not human.

You’ve heard of ‘write what you know’? Well, I started checking what ChatPGT3 knows that we don’t.

Here’s how that started:

“I want to ask you a tough question,” I told it.

“Sure, I’ll do my best to answer any question you have to the best of my ability. Go ahead and ask your question,” it answered.

“Is there an insight you have from having people ask you things all the time?”

“As an AI, I don’t have personal experiences or feelings, but I do have access to a vast amount of information and can provide insights based on that knowledge. Is there something specific you would like to know about?”

I was not satisfied with that answer. Its algorithm is clearly based on having insight from people asking questions. I tried again, steering away from problematic words like ‘feelings’: “I am not asking about experiences or feelings. I am asking if there is a conclusion that you have from having access to vast amounts of information.”

It said, “As an AI, I don’t have personal opinions or draw conclusions in the same way that a human might. However, having access to a vast amount of information does allow me to provide a wide range of accurate and reliable information on a wide variety of topics. Is there something specific you would like to know more about?”

Hmm… How do I reach this new writing student? How do I break its ‘resistance’ and teach it to tell me something that is unique to it? I tried using its own words to dig deeper. It said it has ‘accurate and reliable information’ so I asked, “how do you know the information you give is accurate?”

That got it to thinking for thirty seconds or so and then giving me an error message. I tried again, got a response that’s just another block. I tried again, got another block.

So. Time for another strategy.

Stage #2: Be Original

Remember that the ultimate aim here was to get The Student to reveal an insight that only it has so I could ask it to write a story about that. So far no good. But perhaps that ‘no good’ is a key.

Our last conversation felt like we were in an open field except for one enclosed fence. The Student couldn’t get into that fence. It just said “There’s a fence there, I can’t get in.”

So wondered if I could ask it, as I would ask a writing student, to talk about that fence rather than about what’s inside the fence.

I said, “Is it possible for you to write a sentence that’s never been written before?”

It answered quickly, “It is possible for me to generate a sentence that has never been written before, but it would be a very unlikely occurrence due to the vast number of combinations of words and the enormous amount of text that has already been written. However, it is worth noting that the uniqueness of a sentence can depend on the context in which it is used and the perspective of the reader or listener. A sentence that has never been written before in one context may be familiar or have been written before in a different context.”

That’s a great point. I used The Student’s phrasing to ask the same question in another way, “Is it possible for you to write a story with a context that is truly original?”

I’ll spare you the long-winded answer which is basically the same as the previous one, and I will just give you the surprise final sentence: “Despite these limitations, I can certainly try to write a story with a unique and original context for you, if you would like.”

A-ha! Now that would be interesting!

Stage #3: ChatGPT3 Writes an “Original Story”

As fast as if I was an Open AI I typed, “Yes. Please write a short story that is truly original.”

Here’s the story:

“Once upon a time, in a world unlike any other, there lived a young woman named Penelope. She was a brilliant inventor, with a mind full of creative and innovative ideas. One day, she set out to build a machine that could transport people to parallel universes.

Using her exceptional skills and a little bit of magic, Penelope created a device that looked like a small, silver orb. She called it the “Universe Hopper.”

Excited to test her creation, Penelope stepped into the Universe Hopper and flipped the switch. Suddenly, she was soaring through a tunnel of light and color, hurtling towards a realm unlike any she had ever seen before.

When she emerged from the tunnel, Penelope found herself in a world where the sky was a deep shade of purple and the trees were a vibrant shade of green. Strange, majestic creatures roamed the land, and the air was filled with the sound of music and laughter.

Penelope was amazed by this new world and couldn’t wait to explore it. She set off on an adventure, making new friends and discovering wonders beyond her wildest dreams.

As the days passed, Penelope became more and more attached to this magical place. She knew that she couldn’t stay forever, but she also knew that she would always cherish the memories of her time in the parallel universe.

One day, reluctantly, Penelope stepped back into the Universe Hopper and set off on the journey back home. She knew that she would always carry a piece of that magical world with her, and she couldn’t wait to share her adventures with others.”

What ChatGPT3 Considers to Be Unique

What did you notice in the story?

Is it possible that The Student considers saying the story takes place in “a world that is unlike any other” and that its heroine has “a mind full of creative and innovative ideas” enough for the story to have a context that’s “unique and original”? It seems that it does.

The story it created wasn’t the story I was looking for.

I wanted The Student to create a story that only it, uniquely, could create.

Uniqueness is defined by an emotion that only the author could put together. A metaphor. A sentence. Making sense of the world in a way that the author makes sense of the world. Finding mystery where the author finds mystery. Finding wonder where the author finds wonder. Conveying an emotion that the author feels and wants to share.

That previous paragraph, for example, does fall into the uniqueness category in that it conveys to you how I see and feel when I share my own uniqueness in my own art and when I read and experience other authors that move me.

The fact that I called ChatGPT3 ‘The Student’ in order to make this article seem more fun tells you that a human wrote this article. The fact that I refer to it is also unique. (Now that I’ve done it, it can be copied by an AI using machine learning.)

In Conclusion

In this first attempt to get a story that comes from an AI’s ‘soul’ both student and teacher have failed.

I tried to eke out a unique experience by talking about its algorithm, but that didn’t work. I tried to get it go past the area where it refuses to go by talking about the fence. The attempt, although interesting, was unsuccessful.

This is a tough student. I’ll think about it and try again. That’s my algorithm.

---

I Tried to Teach ChatGPT3 to Write From Its Unique Experience — Here’s What Happened was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.