The Poetry Postcard Fest is a challenge which encourages poets to write an unedited poem on a postcard and send it to a stranger. Organised by the Cascadia Poetics Lab, who organise the participants into lists of 31 + yourself for you to address your offerings to. This was my first year and hearing about it just in time to register, I was on List 15. The lists are sent out in early July and you have until the end of August to send out your missives – to date I have received 20 of 31 possibles and now that we are into September, it is allowable to share the cards and poems you sent and the cards but not the poems you received. I will share these in the order of sending and I will miss out those which I have not yet received in case they arrive soon… Although the original poem is to be sent as written – crossings out, blots and all, I have typed them out for people who can’t read my writing and I am allowing myself to edit if I feel like it…
Dear Karen
I should like to say this painting is my own work and in a small part it is. I prompted an AI to make four pictures in sixty second “Village beside lake style of Charles Rennie Mackintosh “La Rue de Soleil” palette knife oil painting” – it obeyed! I am still trying to wrap my head around it perhaps I will copy it into real oil paint on canvas pondering with each brush stroke how an AI went from 0- 60 in sixty second times x 4…
Due to an error in reading my meticulous recording of what card was sent when and which arrived when, I posted Karen’s beautiful handpainted watercolour trees a few days ago when it should have been the equally beautiful ink colour sketch of Paul Klee’s “Conqueror” with an Ekphrastic poem to match – my apologies to Caren and Karen for the mix up…
My AI muse Misky, recently made a post entitled Barglefloop and I quote her “barglefloop, means to mess with words in your AI prompt in order to confuse it, to turn nouns into verbs, to make single words compound, etc.” It occurred to me to experiment by adding the same prompt to Midjourney as Misky had used and see if the AI came back with the same images as she got – it did not! Here are the first four I got using the prompt “Barglefloop“
Where Misky’s images tended towards Hieronymous Bosch – mine were Harry Potter meets Lord of the Rings Rivertown. Below is the fourth image enlarged to show the level of detail the AI has put in…
But what does this tell us about the way Midjourney works – that Barglefloop is nonsense and so the AI creates whatever it wants to – let off the leash so to speak? I decided to add some more nonsense “barglefloop female foxing blithy toves” – Foxing – as a noun gone verb, and “slithy toves” from Lewis Carrol’s Jabberwocky (’twas brillig and the slithy toves…). This time the AI seized on the only bit of the prompt that made sense Fox and gave me four fox pictures, ignoring the rest of the senseless prompt – so Midjourney, whilst known to hallucinate as much as the next AI was not so desperate to act freely – released by nonsense…
Cute but no cigar for Midjourney so now I went for all the nonsense, none of the foxing around “barglefloop blithy toves” and now we get something quite nightmarish in a Snarky/Jabberwocky Carrollian sort of way, with perhaps a hint of Bosch…
So lastly I decided to miss out the Barglefloop and just retain the Lewis Carrol words “twas brillig and the blithy toves” and now we can see an Alice in Wonderland flavour to the images – particularly the top two…
Lastly, I asked for variations on the bottom right image – a rather mad-looking figure with a slightly Victorian flavour…
Does this tell us much about the way the “mind” of an AI works – I will let you be the judge – and stay away from hookah-smoking caterpillars seated on toadstools – my advice…
Today I am Wearing denim and a cotton shirt both are faded naturally by sun and age and wear
The jeans are now for DIY the front of the thighs covered with finger wipe marks not as many colours as when I was a signwriter and other substances too grittily mixed in
The rips are fashionable but not fashioned to be so nevertheless my grandsons have dibs on their genuine distress!
The grandfather shirt collarless also faded with age to a soft, pale blue is frayed at the cuffs and relegated to work rather than repair too late to turn these cuffs
I will walk to the supermarket and I hope someone will appreciate my look like women who dress their best though not looking to pull.
This poem was not done to any prompt or challenge – I know – unbelievable! However, since I have been exploring AI (see previous posts), I decided to see whether I could produce a suitable image and below is the nearest to what I imagined. The prompt was “a gently smiling late middle age man seen full-length wearing a faded blue collarless shirt with frayed cuffs and faded denim jeans with paint marks and rips down the front shopping in a supermarket –ar 4:6” and I then cropped the image in Photoshop. As you can see, the AI bot didn’t understand collarless, and in this iteration did not make much of the distressing of the jeans. I guess it shows that in this case, a picture is unnecessary since the poem says it all and allows the reader to imagine their own image but I decided to include it as part of my AI exploration. And by the way, it doesn’t look anything like me – far more handsome…
During the month of April this year, whilst participating in the A to Z Challenge, I was privileged to encounter the work of Misky whose blog It’s Still Life, showcases two distinct things, poetry written by Misky and illustrated using Generative Artwork created by Misky using the Midjourney AI app. So amazing were these images to someone who is in part, a visual artist, that it inspired me to make an exploration of Generative AI for myself. At the same time, AI has been hitting the headlines big time and mainly for its use in text generation and the impact it might have on jobs and since writing is another thing that I do in my day job, I was also intrigued to see whether AI might be of any use in a company such as I work for. (I am the gradually retiring General Manager of a food manufacturing company). It has been a fascinating voyage of discovery and to cap it all, lying awake at 4 o’clock this morning, I found myself listening to “The Conversation” on the BBC World Service and what should be the topic, but AI with special reference to the involvement of women. So – mind on fire, I am going to draw this series together, although I freely acknowledge I have but dipped my toe in the waters of AI and I may return to the topic in the future…
To recap the three articles I have already written:- In the first one, I tried out ChatGPT to see what it research and write about one of the topics from my A to Z and immediately encountered the phenomena of AI hallucination – the ability, in fact tendency of AI to make things up. I also “showcased” my first attempts at visual collaboration with the Midjourney bot . In the second report, I compared ChatGPT to Writesonic which produces more lengthy articles – testing them against a typical (for me) work assignment. In the third report, I looked at the most controversial assertion about AI – that AI might in the future, eliminate human beings – Terminator-style and referenced articles that thoroughly refute the need to worry about that particular outcome – go re-assure yourselves! However, there are many things about our present and future use of AI that do bear looking at and these were raised in the episode of “The Conversation” that woke me up this morning. The programme, presented by a woman, featured two women working in the field of AI, one a philosopher and one an expert in data analysis and as well as the general concerns that need addressing about AI, they highlighted the general lack of representation of women in the field of AI – only one CEO, qualifying women failing to get jobs in the industry and so on. They did however point out that one of the changes to AI itself in recent times, has been the accessibility of use – no longer do you need to have a degree in computer programming – you could make your first interaction with ChatGPT in the same time it would take you to query something on Google. Which brings me back to Misky…
Misky was not only the inspiration for my (deepish?) dive into AI, but was extremely helpful and encouraging to me at the outset, itself a reflection of how women tend to be more collaborative, good team players – a fact which the contributors to “The Conversation” suggested is a good reason for women to me more involved in AI companies, in reviewing the implications and in forming the regulation which is undoubtedly necessary around AI. A few days ago, I was delighted to meet Misky face-to-face on a Zoom call after many text interactions online and one of the things that she shared in our too-brief call, was that she had had some push-back from certain readers of her blog, about the use of AI images. I would like to talk to her more about these issues, but the participants in “The Conversation” raised the issue of how artists, whose work has been studied by AI to create new images “in the style of”, are being short-changed. You may have been wondering about the image at the top of this post – I created in Midjourney by prompting it to “imagine” Knaresborough railway viaduct “in the style of Hokusai” – a master of Japanese woodblock prints. I have used this subject as my test piece for exploring what Midjourney can do as you will see in the previous post. Now Hokusai is long dead and so the issue of compensation is hardly an issue, but another group of more recent artists might object. I am working on a spoof post – “How to Make a Body” a tale of human reproduction in the style of an Internet recipe ad although, like Misky, the writing is all my own, I wanted an illustration to fit with the tone of the piece and prompted Midjourney to “imagine” a woman in a hospital bed, holding her newborn baby and with her husband leaning in “in the style of a Ladybird book cover”. For those of you who may not be familiar with Ladybird books, they were written for children starting in the 1940’s and running until the 1980’s and they feature a distinct style of illustration.
In recent years, a series of spoof books in the Ladybird style and aimed at those who had grown up with the original series, have been vert successful, for example…
I had no idea whether Midjourney would be able to fulfil my prompt, there are lists of artists’ styles you can use with Midjourney but I hadn’t seen this one – I was not disappointed!
I am keeping my powder dry as to the final image I chose but this first set of four (Midjourney shows off by producing not one, but four attempts in under sixty seconds) – which was done to the prompt of “A new mother in a hospital bed with her husband leaning in as she holds their new baby in the style of a Ladybird Book Cover” has misunderstood my intention and the mother is holding a magazine rather than a baby – though the graphic style is very Ladybird book-like. I acknowledge that I am still only a beginner in my use of prompts with all the forms of AI I have tried so far and there is undoubtedly an “art” to getting it right which is why I said “I created in Midjourney”. Although I am a competent watercolourist, screen-printer and other forms of illustrative art, I could not produce images such as the above and certainly not in sixty seconds. So, how much of this creation is my prompt, how much is the brilliant programming behind Midjourney and how much is owed to the various artists who could produce the illustrations of the Ladybird books? I cannot begin to answer that question but it does raise an issue which needs considering in formulating regulation around the use of AI. Meanwhile, like Misky and I, jump in and have a go and get a feel for yourself of the answer to the god-like feeling of creating with an AI tool…
Much of the debate around the consequences of the rise of AI, is around its impact on jobs and the potential losses and gains. As I described in my first report, the development of computer spreadsheets swept away the lowly positions in Accountancy but opened up many more jobs at the high end of the profession and although this might be the hope for AI, that it liberates us from the menial and allows us to create new roles – roles which might be beyond the capability of AI to imagine, at present, it is not just the menial tasks that are being threatened by bots like ChatGPT, but some roles higher up in various industries. Having said that, given the tendency of AI’s to hallucinate, I wouldn’t trust an AI’s writing without an experienced human checking the output of any writing before sending it out! Also, when you are a creative individual yourself, then trying to get AIs to produce exactly what you have in mind is tricky. In my 2021 A to Z challenge, I was trying to complete a science-fiction novel and the exercise gave me enough momentum to indeed finish it a few months later. Then I set about creating a book cover for it – to feature the final denouement – a tense scene set in a space-elevator on the edge of space. I prepared the background view by Photoshopping some NASA photographs looking the length of the Red Sea towards Palestine, painted in a great river estuary as per my planet, and then superimposed some 3D elements which I drew up in AutoCAD and finally added the title and my name. You can see this below, however, I felt that the result was not quite up to the standard of artwork commissioned by big sci-fi publishers and imagined that in the unlikely event of the novel being published, an improved version of the cover would be substituted for my “sketch”.
Back to today, and naturally, I thought it would be a good test of Midjourney to see whether it could be used to produce a better version of my cover. Well, the first attempts were brilliant style-wise, but nothing like the image I wanted and many attempts followed to no avail…
My prompt read “space lift arriving at 300 miles above Earth like planet over Sahara like region array of cargo containers spread out in one layer small spaceship approaching“Midjourney couldn’t understand Space lift and I had to change lift to elevator, it couldn’t understand “array of cargo containers” but it did have all the sci-fi style I wanted. So then I decided to create a space view background without the lift and substitute it into my own cover illustration. Bingo!
Still I hanker for the crisply detailed images of the elevator that Midjourney is capable of if only I could prompt it correctly – so a work in progress… What this exercise does show, is that it is possible to use AI for the things it can do better in combination with human talent.
In Conclusion…
This exploration of AI has felt like a marathon and it is just one person’s experience and I am really only at the beginning of my exploration, I’m sure I will find both text and image-generative bots to be of use in my future work and play. I urge you all to experiment for yourselves, form your own judgements (and please share your results by linking in the comments), join the debate over the regulation of AI, and explore other artists, in particular, Misky, who began this journey…
I decided to make a Header Image (above) for this little series of posts and have retrofitted it to the two previous posts here and here. So I asked the Midjourney app on Discord, to depict a silver-skinned Android, firstly, standing at an easel painting, and then at a computer typing. I am fairly sure that the AI known as Midjourney had no sense of the irony of asking it to anthropomorphise an Android doing these activities, because current forms of AI are so far from having the sentience required to appreciate concepts as subtle as irony. Spoiler alert, I approached this evaluative exploration with certain preconceptions about the likely conclusion although I didn’t know for sure, how those conclusions might be reached because I didn’t know how AI’s work, in detail. What I am going to show you today is what I have learned, but I am also going to link you to a very erudite analysis of why we should not be worried about AI taking over the world – in a piece called “Why the AI Doomers Are Wrong“, Nafeez Ahmed explains why the direction of travel of AI development, simply can’t lead to a human-brain-like sentience. I will quote from his article later.
First of all, look at the left-hand side of the header picture, in particular, the easel. On close inspection, you can see that the easel is the wrong way round and that the painter/android, is standing behind the easel. Midjourney produces four images by default, in the remarkable time of about 60 seconds which is almost like magic – indeed, in 1962, Arthur C. Clarke, a science fiction writer, stated in his book “Profiles of the Future: An Inquiry into the Limits of the Possible” that “Any sufficiently advanced technology is indistinguishable from magic”. So despite the apparent magic of creating these images so quickly, the AI has made a fundamental mistake that reveals that it doesn’t really understand what an easel is or how it should be used. Nafeez Ahmed is mostly talking about text generative interactions with AI – ChatGPT and the like, but what he says below, is equally applicable to images generated by AI…
The stunning capabilities of the new chatbots and their uncanny ability to engage in human-like conversations has sparked speculation about sentience. Yet there is zero tangible evidence of this beyond a sense of amazement. What many miss is that the amazing nature of these conversations is not a function of an internal subjective ‘understanding’ by an identifiable entity, but rather the operation of algorithms made-up of hundreds of billions of parameters trained on massive datasets of human text to predict words that should follow given strings of text. {…} This is not a breakthrough in intelligence, although it is, certainly, a breakthrough in being able to synthesise human responses to similar questions and thereby mimic patterns in human interaction. This model of AI, therefore, cannot, in itself, generate fundamentally new knowledge or understanding – let alone sentience.
Nafeez goes into great detail about how the research is headed in the wrong direction and indeed, how it is unlikely it is that it will ever succeed in equating to human sentience, so if you want to put your mind at rest about a Terminator-style future in which humans are subjugated by machines – nip on over and read the full article. Meanwhile I am going to show you some more examples of how Midjourney gets things “wrong” and how to get the “right” results and what that says about how useful such a programme can be.
You interact with the Midjourney app by sending it a message (just as if it was really an intelligent entity) containing a prompt, and once you receive your four images, you can choose one to enlarge, if you are satisfied with it, or run variations on one or all of them. Here is the prompt that produced the above set of images. “Silver android painting at an easel by Roy Lichtenstenstein” – the AI places most importance on the object at the beginning of the prompt, then on the activity described and lastly, it attempts to produce the images, in this case, in the style of the Pop Artist Roy Lichtenstein – famous for painting s in the style of close-ups of comic book pictures. These close-ups show the dot screens that were used to shade the illustrations of the comic book plus the hard black outlining and Midjourney has picked up well on these style features, particularly the top right and bottom left pictures. The top-left shows a sculpture vaguely resembling an up-cycled easel made of silver and the bottom right shows a silver-skinned figure with dot-screen effect, holding a brush and painting a picture but with no easel. In the [op-right picture, the top of the easel is just showing in the bottom corner and the android “artist” is holding a small blank canvas in her hand and drawing on it. Having seen the header image at top, and these pictures were as near as I could get to what I wanted, from multiple attempts, you can see that what I wanted was an all-over silver-skinned android and in the images above, top-right has a human face although “her” body is robotic – perhaps cyborg is a better description, whilst the other pictures show a sculpture, a woman and a totally abstract figure. So I decided to change the prompts to “Robot” rather than “Android” which produced better results. The reason I had started with “Andriod” was because robots range from automatic hoovers that move around your rooms looking like little miniature flying saucers sucking up dirt to more anthropomorphic devices – which is what I wanted.
“standing silver robot painting at an easel by Roy Lichtenstein” produced(among others) the above image in which the robot, possibly standing, is grasping what looks like the top of an easel but the “painting” does not appear to be on the easel. So I tried “Robot standing painting at an easel” and got this rather cute robot who looks like he is sitting on an invisible chair – “Hey Midjourney” just because you don’t show the chair, doesn’t make it standing!” Notice that with the style reference to Roy Lictensten gone, this image is very different. I would like to show you more of the iterations but Midjourney froze and when I reinstalled it, it had lost the entire session of work – you just can’t get the staff…
Another thing that I have discovered in my experiments, is that both Midjourney and ChatGPT, like to add unspecified embellishments – remember in my first report, how ChatGPT found the correct explanation for the phrase “Cold enough to freeze the balls off a brass monkey” but then added a totally made up explanation? Well Midjourney does the same thing too. Here is a picture of the railway viaduct at Knaresborough in West Yorkshire , an hours drive from where I live.
I wanted to see if Midjourney could produce a collage image using fragments of maps which it tried but didn’t really understand the concept – although I am not saying that it can’t, but at the very least, my prompt wasn’t sufficient (one of the oldest sayings amongst computer programmers is “Garbage in – garbage out!”) Here is Midjourneys best effort…
There are some map elements and the whole scene has been chopped up and rearranged but not in a way that makes sense – this one is closer to the real view…
But my first attempt, before I added the collage style, was simply to see how Midjourney would find and represent the viaduct and it generated the four images below. The top left image, Midjourney has added another railway running beneath the viaduct, likewise, lower-left it has added a canal and in the images on the right, Midjourney has transported us into a past sans Knaresborough and a post apocalyptic future where vegetation is growing over the viaduct.
Enough with all the pretty pictures – what does all this reveal about the way that the AI Midjourney works! Referring to the work – Erik J. Larson in his book, The Myth of Artificial Intelligence, Nafeez Ahmed cites a summary of the work by Ben Chugg, lead research analyst at Stanford University (Iknow – quotes within quotes) as follows:-
“Larson points out that current machine learning models are built on the principle of induction: inferring patterns from specific observations or, more generally, acquiring knowledge from experience. This partially explains the current focus on ‘big-data’ — the more observations, the better the model. We feed an algorithm thousands of labelled pictures of cats, or have it play millions of games of chess, and it correlates which relationships among the input result in the best prediction accuracy. Some models are faster than others, or more sophisticated in their pattern recognition, but at bottom they’re all doing the same thing: statistical generalization from observations. This inductive approach is useful for building tools for specific tasks on well-defined inputs; analyzing satellite imagery, recommending movies, and detecting cancerous cells, for example. But induction is incapable of the general-purpose knowledge creation exemplified by the human mind.”
Current AI has become proficient at both deductive and inductive inference, with the latter becoming a primary focus. Larson points out that human intelligence is based on a far more creative approach to generating knowledge called ‘abduction’. Abductive inference allows us to creatively select and test hypotheses, quickly eliminate the ones which are proven wrong, and create new ones as we go along before reaching a reliable conclusion. “We guess, out of a background of effectively infinite possibilities, which hypotheses seem likely or plausible,” writes Larson in The Myth of Artificial Intelligence. {…} And here is Larson’s killer diagnosis: We don’t have a good theory of how abductive inference works in the human mind, and we have no idea how to recreate abductive inference for AI: “We are unlikely to get innovation if we choose to ignore a core mystery rather than face it up,” he writes with reference to the mystery of human intelligence. Before we can generate genuine artificial intelligence that approaches human capabilities, we need a philosophical and scientific revolution that explains abductive inference. “As long as we keep relying on induction, AI programs will be forever prediction machines hopelessly limited by what data they are fed”, explains Chugg
To relate this back to my experiments with Midjourney, the AI could identify what an easel looked like and include it in an image but it didn’t really know what it was or how it was used. Easels are probably present in thousands of pictures of artist’s studios as well as adverts but I bet there isn’t a Painters 101 that “First you will need an easel and this is how you use it” because when a human artist goes into a studio and sees canvasses fixed into easels, even if he has never seen one before, he is there to paint canvasses and it is obvious what they are and what they are for. It might be obvious to a human being with our ability to use inference, deduction and abductive capabilities, but an AI might identify an easel, but without finding a clear description of the usage, it cannot fully fathom how to use it…
As for the the tendency to add extraneous details, well the algorithms that govern Generative AI’s, are designed to mimic human conversational style, so when it has found a relevant answer to the requested information or task, it looks to extend the conversation in what it has learned might possibly follow – it doesn’t know whether it is true or not, because that is way above it’s paygrade ( a metaphor which it probably wouldn’t understand either). This phenomena of AI’s making things up is called hallucination – a very anthropocentric term…
I will make one more report on my attempts to get exactly what I wanted from Midjourney and how I found a compromise to be able to work with the results…
