Category: Artificial Intelligence

TL:DR – I used ChatGPT to pick five of the best pieces of advice from a list of 24 offered by Werner Herzog.

I was always intrigued by the idea of Werner Herzog, although I must confess I’ve only seen a handful of his films, among perhaps the most well renowned, Fitzcarraldo. Anyway, an article passed across my social media timeline today listing 24 pieces of advice from Herzog. Some might apply to all of us, some are rather figurative, others are more pertinent to filmmaking, but could easily apply to other areas of human endeavour. Anyway, 24 seemed like too many aphorisms at the time, it was a Monday morning, and so I turned to ChatGPT and asked it to select just five from the 24 that might be seen as universal advice, lifehacks, for all of us.

I asked it for its reasons for picking those over the other 19 and it basically replied that each of the others could be seen as being more pertinent to filmmaking and not particularly relevant for non-filmmakers. I don’t think that was entirely true as each of the others could be taken as advice in other contexts, but the top 5 are pretty good advice for us all. Here is the list of Herzog’s original 24 pieces of advice.

TL:DR – I asked ChatGPT, the language model I’ve been talking about a lot recently, to explain itself in simple terms, which it did.

I want you to tell me about ChatGPT but you must assume I have no specialist understanding of any jargon terms like “AI”, “language model”, “algorithm”, so your response must explain those terms if they are used in your explanation.

ChatGPT is a type of Artificial Intelligence (AI) program that is specifically trained to understand and generate human language. It uses a type of machine learning called deep learning, which involves training a computer to learn patterns in large amounts of data, such as text.

In simple terms, ChatGPT is a computer program that can understand and respond to text in a way that mimics human conversation. ChatGPT uses a type of machine learning called deep learning, which allows the computer to learn from a large amount of data. It’s like a virtual assistant that can answer questions, write sentences, and even compose stories. It can understand and generate text in a way that’s almost like talking to a person.

I understand most of that but what do you mean by the phrase “learn patterns in large amounts of data” in what way is the AI “learning” patterns and what do you mean by patterns?

When we say that ChatGPT “learns patterns in large amounts of data,” we mean that the AI is analyzing and identifying common themes, structures, and relationships in the data it is trained on.

For example, let’s say ChatGPT was trained on a dataset of written conversations between people. As it reads through these conversations, it begins to identify patterns in how certain words and phrases are used in different contexts. It might notice, for example, that the word “because” is often used to connect two related ideas, or that the phrase “I’m sorry” is often used as an apology.

These patterns that the AI identifies are called statistical patterns, they are regularities or repetition of certain elements, words, or structures in the data. These patterns are then used by the AI to make predictions or decisions.

When we say “learning” in the context of AI, we mean that the computer is using these patterns to make predictions or decisions. In the case of ChatGPT, it’s using the patterns it learned to generate text that is similar to the text it was trained on.

In summary, “learning patterns in large amounts of data” means that ChatGPT is analyzing data, identifying statistical regularities, and using them to make predictions or decisions.

So, how many patterns has ChatGPT learned?

It’s difficult to say exactly how many patterns ChatGPT has learned, as the AI continues to “learn” new patterns even after it has been trained. However, the number of patterns that ChatGPT has learned can be estimated by the size of the dataset it was trained on and the complexity of the task it is trained to perform.

ChatGPT is trained on a very large dataset of text, which contains billions of words and phrases. It has been trained to generate text in a wide variety of styles and formats, such as conversation, news articles, fiction, etc. With this amount of data and complexity, ChatGPT has learned a vast number of patterns and regularities in the use of language, and it continues to learn as it is exposed to new data.

It’s worth noting that the language and the way it’s used is very complex, so the number of patterns that ChatGPT has learned is also complex and hard to quantify.

Okay, but give me a back-of-the-envelope calculation to say approximately how many patterns ChatGPT has learned.

I can give you a rough estimate of how many patterns ChatGPT has learned. For example, let’s assume that a “pattern” is defined as a unique word or phrase that appears in the dataset. With a dataset that contains billions of words, it’s likely that ChatGPT has learned several million unique words and phrases.

However, this is just a rough estimate, and it doesn’t take into account the many different ways that words and phrases can be used in different contexts, which would require a much more complex analysis. Additionally, It’s also important to note that ChatGPT doesn’t only learn patterns based on single words, but also based on word combinations, phrases, idiomatic expressions and also the context in which they appear.

In summary, it’s hard to give an exact number for the number of patterns that ChatGPT has learned, but it’s likely that the number is in the millions.

What will happen inside the ChatGPT system when it has been trained on all the text in the world and learned all the possible patterns in that data, will it understand more than a human.

It’s unlikely that ChatGPT, or any AI system, will ever be trained on “all the text in the world” as the amount of text is constantly growing and changing. Additionally, the process of understanding text goes beyond just recognizing patterns in the data, and is also influenced by factors such as context, intention, and personal experience.

Even if ChatGPT were to be trained on a very large dataset of text, it’s unlikely that it would understand text in the same way that a human does. While ChatGPT can recognize patterns in the data it was trained on and generate text that is similar to the text it was trained on, it doesn’t have the same level of understanding of the meaning and context of the text as a human does.

Furthermore, understanding human language is a very complex task, as it goes beyond just recognizing patterns in the data, it also involves understanding the context and the meaning behind the text. For example, ChatGPT may not be able to understand sarcasm, idiomatic expressions, or jokes, it may also not be able to understand the emotions and intentions behind the text.

In summary, while ChatGPT can learn patterns in large amounts of data, it doesn’t have the same level of understanding as a human, and it’s unlikely that it will ever achieve a human-like understanding of language.

How does ChatGPT know how to respond to a prompt if all it is doing is learning patterns and then outputting words that follow the appropriate pattern when prompted.

ChatGPT uses a technique called “conditional generation” to respond to a prompt. When it is given a prompt, it first analyzes the input to identify the context and any relevant information. Then, it uses the patterns it learned during its training to generate a response that is appropriate to the context of the prompt.

The way it works is that ChatGPT uses the patterns and regularities it learned from the training data to generate text that is similar to the text it was trained on. For example, if the prompt is a question, ChatGPT will use the patterns it learned to generate an answer that is similar to the answers it saw in the training data.

It’s important to note that ChatGPT does not have a “memory” in the traditional sense, it doesn’t “remember” specific inputs or responses from the past. Instead, it generates responses based on the patterns it learned from the training data.

It also uses a technique called “beam search” to generate its responses, which means it analyzes multiple possible responses and chooses the one that is the most likely, based on the patterns and regularities it learned.

In summary, ChatGPT uses the patterns it learned during its training to generate text that is similar to the text it was trained on, and the response is generated by analyzing multiple possible responses and choosing the one that is the most likely, based on the patterns and regularities it learned.

Is there anything else I need to know about how ChatGPT works?

Here are a few additional things to know about how ChatGPT works:

Pretraining: ChatGPT is a “pretrained” model, which means that it was trained on a large dataset of text before it was fine-tuned for specific tasks. This allows it to generate text that is similar to the text it was trained on, even when it is given a new prompt that it has never seen before.

Fine-tuning: Once the model is pretrained, it can be fine-tuned for specific tasks or domains. This means that it is trained on a smaller dataset of text that is specific to the task or domain. Fine-tuning allows the model to generate text that is more relevant to the specific task or domain.

Limitations: While ChatGPT is a powerful language model, it still has limitations. It is not able to understand the meaning and context of the text in the same way that a human does, and it may generate text that is not accurate or appropriate.

Human supervision: ChatGPT’s generated text should always be reviewed by humans to ensure that it is accurate and appropriate before it’s used in any application.

Continuous learning: ChatGPT’s learning is continuous, this means that it can learn new patterns and regularities as it is exposed to more data, this allows the model to adapt and improve over time.

In summary, ChatGPT is a powerful language model that uses deep learning to generate text that is similar to the text it was trained on, but it still has limitations. It’s important that its output is reviewed by humans to ensure accuracy and appropriateness before it’s used in any application. Additionally, it can continuously learn new patterns and regularities as it is exposed to more data.

TL:DR – You can get ChatGPT to analyse a piece of prose for style and tone and then use the results to get it to generate a new piece of prose on another topic with an approximation of that style and tone.

If you feed ChatGPT, the AI language model, an appropriate prompt followed by a sample of writing, it does something that could be rather useful. The bot analyses the writing style. The prompt needs to be something like this: “Analyze the text below for style, voice, and tone. Then, using natural language processing (NLP) create a prompt to write a new article in the same style, voice, and tone.” and you paste in a sample of your own writing.

I gave the bot this prompt and pasted the text from a recent blog post of mine about migratory blackcaps in our garden. It came back with this summary:

Style: Informal, observational
Voice: First person narrator
Tone: Thoughtful, contemplative

I’d agree with that, I certainly try to write my nature posts in that kind of style, voice, and tone. The bot also generates a new prompt based on its analysis:

Prompt: Write an article about observing a different species of bird in your backyard and the possible reasons for changes in their migration patterns. Use personal anecdotes and scientific explanation to convey a thoughtful, contemplative tone.

This prompt can then be used to generate a new article. So, I tried that but I modified the prompt to focus on a particular species of bird, the Fieldfare, rather than using the phrase “different species of bird” and I changed the word backyard to garden. This is the article that ChatGPT generated, you can compare it with my own original article on the topic of Fieldfares in the garden, which wrote a few years ago for the blog.

NB The following article was generated by the AI to emulate my writing style!

Fieldfares in the Garden: A Wintertime Delight

As a nature lover, I have always enjoyed watching the various birds that visit my back garden throughout the year. Recently, I have had the pleasure of observing a type of thrush known as a Fieldfare. These birds are typically found in Northern Europe and Asia, and are known for their striking plumage, consisting of a grey head and mantle, with a chestnut-colored breast and belly.

I first noticed some Fieldfares in my garden during a harsh winter period. They were attracted to the berry bushes in my garden. Their presence was a welcome sight, as the dull grey days of winter can sometimes feel quite monotonous.

One possible explanation for change [in their migration patterns] could be the milder winter temperatures that have been observed in recent years. Fieldfares have a high metabolic rate and require a lot of food to survive the cold winter months. Milder temperatures may make it possible for them to find enough food in their breeding grounds, reducing the need for them to migrate as far south.

Another potential explanation could be the availability of food in urban areas like my garden. As more and more people take up gardening, the amount of food available to birds in urban areas has increased, making it less necessary for them to travel as far to find sustenance. This may also explain why I have observed more Fieldfares in my garden over the years, as the berry bushes have matured and now provide a reliable source of food.

As I watch the Fieldfares flitting from bush to bush in my garden, I am reminded that the natural world is ever-changing, and that even small changes in the environment can have a big impact on the lives of the animals that call it home. It is important that we continue to observe and study these changes, so that we can understand and appreciate the intricate connections that exist within our natural world.

Note, there were a couple of sentences that were factually incorrect regarding the numbers of Fieldfares I have actually seen in my garden and some aspects of their behaviour which did not ring true, so I have removed those. I also asked the AI to give the article a title.

TL:DR – I asked the language model ChatGPT, to give me an answer to the question: “Is it unethical to use ChatGPT to write a blog post?” The following “chat” ensued.

It is generally considered ethical to use language models like chatbot to generate text as long as you disclose that you used a language model and do not present the generated text as your own original work. However, it is important to use your own judgement and consider the specific context in which you are using the language model. If you are using the language model to generate content for a blog post, it would be a good idea to clearly disclose that you used a language model and to provide appropriate attribution. This will ensure that readers are aware of the source of the content and can make their own decisions about its value. It is also important to consider whether the use of a language model is appropriate for the specific purpose and audience of your blog post.

I then suggested that I didn’t think most people would consider it ethical. It answered as follows, reiterating what it had already said and expanding on the topic a little.

There is some debate about the ethics of using language models to generate text, and different people may have different views on the matter. Some may argue that it is acceptable to use language models to generate text as long as it is clearly disclosed and properly attributed, while others may have concerns about the potential for language models to be used to deceive or mislead readers. Ultimately, it is up to individuals to decide what they consider to be ethical in a given context. It is always a good idea to be transparent about the sources of your content and to give appropriate credit to any sources that you use. This will help to ensure that readers are able to make informed decisions about the content they are reading.

I wanted the AI to expand further and asked why do people think it’s unethical to use AI as it’s just a tool.

There are a few reasons why some people may consider the use of language models to generate text to be unethical. One reason is that using a language model to generate text may be seen as a form of plagiarism, since the text is not originally created by the person presenting it. Additionally, there is a concern that language models could be used to deceive or mislead readers by generating text that is difficult to distinguish from human-written content. Some people may also have concerns about the potential for language models to be used to automate the creation of content, potentially leading to the displacement of human writers. These are all valid concerns, and it is important to be transparent about the use of language models and to use them responsibly.

I questioned its suggestion that the use of such text might be considered plagiarism given that there is no creator to claim ownership.

Plagiarism is generally defined as the act of presenting the work of others as your own, without proper attribution. While it is true that a language model does not have an “author” in the same way that a human does, the use of a language model to generate text could still be considered plagiarism if the generated text is presented as the original work of the person using the language model, without proper attribution. In order to avoid the appearance of plagiarism, it is important to clearly disclose that the text was generated using a language model and to provide appropriate attribution to the source of the text. This will ensure that readers are aware of the source of the content and can make their own decisions about its value.

I think I’ve reached a dead-end when I pointed out that the AI app is essentially just a sophisticated editing tool. In a parallel activity, if I were to convert my photograph into a cartoon using a graphics editor with an AI plugin for making such a conversion, the output would still be a derivative of my original work. Similarly, my language prompt given to the chat AI is analogous to my photograph and its output is simply a derivative work. ChatGPT simply reiterates what it had already said.

It is true that a language model like ChatGPT can be used as a tool to help generate text, much like an editing tool or graphics editor can be used to manipulate or transform a photograph. However, it is important to consider the specific context in which the text generated by the language model is being used. If you are using the language model to generate content for a blog post or other publication, it would be important to clearly disclose that you used a language model to generate the text and to provide appropriate attribution. This will help to ensure that readers are aware of the source of the content and can make their own decisions about its value. Ultimately, it is up to the individual to use their own judgement and to consider the specific context in which they are using the language model to determine whether it is appropriate to present the generated text as their own original work.

TL:DR – The short answer is yes and no, depending on how one uses it, AI can be a prompt for ideas and those ideas might support or stifle creativity.

If you read the previous two Sciencebase articles, you will note that the first one about our New Year trip to Norfolk was a conventional article in my usual style, descriptive, with a few puns and pops, and a load of bird photos. The next article was a reboot of that same article where I fed each paragraph from the former to an AI tool known as ChatGPT and provided its responses.

They were almost insightful, the AI responded in a positive way to my paragraphs basically rewriting them as a response in a way a chatshow host might reiterate and re-emphasise the points made by a guest on the show. It also seemed to extrapolate and extend on open-ended points I made in my original article to make suggestions. For example, I’d used the phrase “for starters” to allude to their being a main course when it came to our birdwatching over the three days. I think the AI picked up in this and while mentioning some birds then asked whether I’d seen any more.

Anyway, I started a fresh chat with the AI bot after writing up that latter call-and-response article and asked it the following question:

“I am wondering how important AI will be in the near future will it boost our collective creativity or stifle innovation?”

It came back with this response, which sums things up in a positive and succinct way, I must admit:

“AI has the potential to enhance and augment human creativity and innovation in many ways. For example, AI can help people generate new ideas by providing them with a diverse set of options or by suggesting connections between seemingly unrelated concepts. AI can also help people evaluate and refine their ideas by providing them with data-driven insights or by helping them identify potential problems or challenges.

At the same time, it is important to recognize that AI is only a tool, and its impact on creativity and innovation will depend on how it is used. If AI is used in a way that complements and enhances human capabilities, it has the potential to boost creativity and innovation. However, if AI is used to replace human decision-making or to automate creative processes, it may stifle innovation. Ultimately, the key to maximizing the benefits of AI for creativity and innovation is to find the right balance between using AI to augment human capabilities and preserving the unique strengths and creativity of humans.“

TL:DR – The language model ChatGPT seems to put a positive spin on everything, even rather negative topics.

I fed the paragraphs from my recent blog post about our new year trip to the Norfolk coast sequentially to the ChatGPT, language model to see what sort of things it would say in response to my words…its responses put a positive spin on everything, even my complaint about the number of dogs!

It’s also interesting, that it recognises that I said “for starters” and so asked if I saw any other interesting birds. For the most part, it looks like the bot repeats back to you what you’ve said like a vaguely interested and patronising friend might! Be interesting to know from where it grabs the additional information about each species and whether that’s copyright material or open to use because it was my creative input that generated it using the app…

I discuss this artificial discussion in an article entitled “Will AI help us be more creative and innovate?”

Here’s the call-and-response chat I had with the AI:

Sciencebase: As has been our habit for the last few years, we have eschewed the midnight festivities of New Year and escaped to the coast. This time, we straddled the New Year with three nights in a cosy cottage in Wells-next-the-Sea. We enjoyed the local hostelries in the evenings during our trip, but the main focus was to walk as far as we could manage each day (usually 7 or 8 miles) and to take in the birding and other sites of nature en route.

ChatGPT: Sounds like you had a great time! Wells-next-the-Sea is a beautiful location and it’s always nice to take a break from the hustle and bustle of everyday life. Walking and birdwatching are both excellent ways to relax and connect with nature. It’s great that you were able to spend some time doing activities that you enjoy and getting some exercise at the same time. I hope you had a wonderful trip and that you’re feeling refreshed and rejuvenated for the new year.

Sciencebase: Wells, Warham, Titchwell, and Holkham Gap were the main areas, beach, woodland, marsh, and nature reserve. A couple of Muntjac (one deceased), two or three Grey Squirrel, half a dozen Grey Seal, and 1001 dalmations and others dogs were the limited list of mammals we saw.

ChatGPT: It sounds like you had a great time exploring the different environments in Wells-next-the-Sea and the surrounding areas. Muntjac, grey squirrel, and grey seal are all interesting animals to see, and it’s great that you were able to spot them during your walks. The variety of dogs you saw must have been fun to watch as well. I hope you were able to get some good photos or memories of your adventures!

Sciencebase: The birding list was much better as you’d expect, for starters, we saw a rare vagrant over the marsh at Warham east of Wells, a Pallid Harrier, along with a couple of Hen Harriers on the same marsh. It was dull and grey at that point and the low-light photos of the harriers are just my record shots.

ChatGPT: Seeing a Pallid Harrier and a couple of Hen Harriers is quite a treat for birdwatchers! These birds are not often seen, especially the Pallid Harrier, which is a rare vagrant to the UK. It sounds like you were lucky to spot them during your trip. It’s understandable that the photos may not have turned out well in the low light, but I’m sure the experience of seeing these birds in person was well worth it. Did you see any other interesting birds during your trip?

Sciencebase: The Pallid Harrier, Circus macrourus, is a migrant that breeds in Eastern Europe, Iran, and central Asia, wintering in India or Africa depending on its migratory wont. It is rare in Western Europe and the UK, but occasional vagrants are ticked here. That said, the species is now known to have bred in The Netherlands (2017) and Spain (2019). A changing world means a changing world for the birds too.

ChatGPT: Yes, that’s correct. The Pallid Harrier is a species that typically breeds in Eastern Europe, Iran, and Central Asia, and winters in India or Africa. It is not a common sight in Western Europe, including the UK, but as you mentioned, there have been a few instances of the species being spotted as a vagrant in these regions. It’s interesting to see how changing environments and conditions can affect the distribution and migration patterns of different species of birds. Did you see any other unusual or rare birds during your trip?

Sciencebase: We saw dozens of other species (around 80). Many of those we had seen before, some many times, but one was rather special and we’d only seen it once before, in Poole Harbour in the autumn of 2022 – White-tailed Eagle. An immature specimen flew over us as we were heading back along the beach to Holkham Gap from the westward marsh end of the patch. The bird itself was heading to its roost on the marsh where it had been reported at roughly the same time for the previous couple of days. We don’t know at this point whether the bird we saw was one of the two we saw in Poole, these Isle of Wight reintroduction birds do cover a lot of ground on their travels.

ChatGPT: Seeing a White-tailed Eagle is definitely a special and memorable experience! These birds are not commonly seen in the UK, so it’s great that you were able to spot one during your trip. It’s interesting that you saw one in Poole Harbour in the autumn of 2022 and then saw a potentially different one in Wells-next-the-Sea a few months later. It’s possible that the one you saw in Poole was one of the Isle of Wight reintroduction birds, as you mentioned. These birds are known to cover a lot of ground on their travels, so it’s not uncommon for them to be spotted in different locations. It’s wonderful that you were able to see such a beautiful and impressive bird during your trip.

Sciencebase: On a smaller scale, but much more numerous, we had some lovely views of visitors from The Arctic, Snow Buntings (30+), which are distant cousins of the Yellowhammers and Reed Buntings. We also saw Shorelarks (about 11) at Holkham Gap despite the best efforts of uncontrolled dog walkers to repeatedly scare the birds away.

ChatGPT: Snow Buntings and Shorelarks are both interesting and attractive birds to see. It’s great that you were able to spot a large number of Snow Buntings and a few Shorelarks during your trip. It’s unfortunate that the Shorelarks were being disturbed by uncontrolled dogs, but it’s good that you were still able to get some views of them. These arctic visitors can be a real treat for birdwatchers, and it sounds like you had a great time spotting them during your walks.

Sciencebase: Below is the, hopefully complete, list of birds we saw, we may have a few others that we may have glimpsed in passing but are not claiming for the list, Grey Partridge, Bullfinch, Whooper Swan, Sparrowhawk. There were no feeders at the RSPB Titchwell cafe area on this visit, so no sighting of Coal Tit on this visit.

ChatGPT: Here is the list of birds you saw, with a few notes about each species:

It sounds like you had a good variety of birds on your list, and it’s a shame that you weren’t able to spot a Coal Tit on this trip. I hope you had a great time birdwatching and that you were able to get some good views of the birds you saw.Sciencebase: I listed all the birds we saw in my original blog post and then fed them to the bot to see what it would give me.

ChatGPT: It looks like you saw a wide variety of birds during your trip! It’s impressive to see such a long list of different species. Some of the more interesting or notable birds on the list include the Pallid Harrier (which is a rare vagrant to the UK), the Hen Harrier (which is a threatened species in the UK), the Great White Egret (which is a large and striking bird), the Red Kite (which has made a comeback in the UK after being nearly extinct in the 19th century), and the Rough-legged Buzzard (which is a rare winter visitor to the UK). It’s great that you were able to see so many different birds and that you had such a successful trip. I hope you enjoyed your birdwatching experience and that you were able to get some good views of the birds you saw.

Sciencebase: I then fed it a shorter list of some of the birds – Avocet, Bar-tailed Godwit, Black-tailed Godwit, Brent Goose, Pink-footed Goose, Pintail, Red-breasted Merganser Red-throated Diver, Shore Lark.

ChatGPT: Here is some information about the bird species you listed:

These are all interesting and attractive birds to see, and it’s great that you were able to spot them during your trip. I hope you enjoyed your birdwatching experience and that you were able to get some good views of these species.

I asked a “plot generator” website to come up with a story based on a couple of dozen words I input. This is what it came up with:

The sleet that rained like chatting lizards

Matt Trescothik was thinking about Jenny Slaughden again. Jenny was a kind juggler with dirty lips and ugly arms. Matt walked over to the window and reflected on his picturesque surroundings. He had always hated damp Wivenhoe with its energetic, excited estuaries. It was a place that encouraged his tendency to feel puzzled.

Then he saw something in the distance, or rather someone. It was the a kind figure of Jenny Slaughden. Matt gulped. He glanced at his own reflection. He was a deranged, splendid, wine drinker with hairy lips and ample arms. His friends saw him as a bulbous, bewildered brute. Once, he had even rescued a silent old man from a burning building.

But not even a deranged person who had once rescued a silent old man from a burning building, was prepared for what Jenny had in store today.

The sleet rained like chatting lizards, making Matt unstable. Matt grabbed a cursed hat that had been strewn nearby; he massaged it with his fingers. As Matt stepped outside and Jenny came closer, he could see the good smile on her face. Jenny gazed with the affection of 5729 intelligent damaged dogs. She said, in hushed tones, “I love you and I want work.”

Matt looked back, even more unstable and still fingering the cursed hat. “Jenny, I like your toes,” he replied. They looked at each other with ambivalent feelings, like two mighty, magnificent mice rampaging at a very greedy accident, which had jazz music playing in the background and two bold uncles sitting to the beat.

Matt studied Jenny’s dirty lips and ugly arms. Eventually, he took a deep breath. “I’m sorry,” began Matt in apologetic tones, “but I don’t feel the same way, and I never will. I just don’t love you Jenny.” Jenny looked barmy, her emotions raw like a grisly, gloopy guitar.

Matt could actually hear Jenny’s emotions shatter into 5009 pieces. Then the kind juggler hurried away into the distance. Not even a glass of wine would calm Matt’s nerves tonight.

THE END

The site also generates reviews (auto praise) of your “story”