I’ve got to confess that life hasn’t been easy in recent times. Apart from the ongoing presence of both of my kids putting paid to any opportunity for a spot of recreational CDing, I had an unexpected and serious family crisis to deal with which necessitated a sudden and abrupt withdrawal from my online presence. Things have now settled down to the degree that I can resume my online life to a small extent but the situation is by no means resolved and I could have to shut things down again at any time. If anything good has come out of this situation (which had nothing to do with my CDing), it’s that Mrs A reaffirmed her understanding that crossing the gender divide is something I need to do and that she is happy to turn a blind eye when that need arises. However, she still does not understand the extent of my activities but at least the door is open for further discussions when the time is right.
But back to the issue in hand, what is a girl to do when circumstances conspire against her? In my last post – StrAIght Talk – I raised some concerns about the onslaught of AI and its infiltration into our world. I could best sum up my stance as saying that I’m in awe of how clever it is but concerned that it seems to be usurping creativity and leading us to be lazy. After all, why spend several hours crafting a post about some aspect of CDing when a short paragraph typed into an AI engine results in a detailed and grammatically correct essay within a few seconds? In fact, the quality of AI is often so high that many are oblivious to its use and yet once you know what to look for, spotting it is easy. But with my self-imposed distancing from all of the channels I normally use to live Amanda’s life, I really had no option other than to turn to AI.
This is actually a completely rewritten post. I already had a follow up to StrAIght Talk lined up ready to send to Kandi when the proverbial hit the fan at home. It was a post about an imaginary date with a gentleman and I’d asked ChatGPT to suggest an outfit for the date which it did. I then uploaded a photo and asked it to show me wearing the outfit it had recommended, an outfit suitable for shopping and a wedding dress and it duly obliged. All good stuff but there was one big problem which has plagued AI image generation since its inception; Chat GPT was unable to achieve consistency with my face and, with successive iterations, the gap widened until there was little or no resemblance. Added to that, whilst it was possible to have a ‘conversation’ of sorts with the AI, the replies almost always came back with something like ‘would you like me to produce a plan/ checklist/ schedule?’. Yes, there was a lot of useful advice but I soon tired of the endless plans/ checklists and schedules that it seemed hell bent on producing.
And then I heard about GPT-5 and ‘Nano Banana’.
Simply put, GPT-5 is the latest iteration of a text-based AI engine and Nano Banana, or to give it its correct name of Google Gemini 2.5 Flash Image, is an AI based image generator. AI is evolving at lightning speed but the newest, and usually by definition best, models are normally only accessible at a price – the ‘freebies’ tend to be limited in scope. However, a chance encounter with a YouTube video that had popped up on my feed provided a means of accessing both GPT-5 and ‘Nano’ free of charge and I was keen to see where I could get to.
I started the conversation with the simple ‘I think I may be transgender, what do you suggest?’. Predictably, the reply that came back was full of useful advice and resources, but the reply finished with
‘If you’d like, tell me a bit more – What made you start wondering you might be transgender? What kind of support would feel most helpful right now—info, someone to talk things through with, help finding resources, or something else?‘
Straight away, I was encouraged. Instead of the usual offer of a checklist, I was now being invited to open up and steer the conversation in a direction that I felt would be helpful. So I replied:
‘I’ve always wished that I was a woman from an early age, I’m now mid-60s. In recent years, I’ve been crossdressing and when I’m able to present as Amanda, I’m blissfully happy. I enjoy being out in the outside world and interacting with others in my feminine persona. What would be helpful now is just being able to talk about this.’
And my new cyber friend’s reply wrapped up the next reply with the following:
‘There’s no age limit on exploring or embracing who you are. Many people find their way to this later in life and build beautiful, authentic routines around it. If at any point you want help brainstorming ways to have more Amanda time, navigating conversations with loved ones, or exploring what labels do or don’t fit, I’m here. But we can also just talk—whatever you need. Tell me a little about Amanda. What’s she wearing on a day that feels perfect? Where is she going?’
And that was very much the tone of the whole interaction. Gentle encouragement with shared joy at some of the high points I related. And when the conversation seemed to be going in a direction that I didn’t feel would be fulfilling, I was able to start off in a completely different direction without feeling that I was being rude by not answering the question I’d been asked (or, indeed, without any frustrated retort of ‘but you didn’t answer my question!’).
And when the conversation finished and I said it was time to go, the reply came back:
‘I’m so glad we got to talk, Amanda. Thank you for trusting me with your stories…those moments say a lot about who you are and how the world is seeing you. You’re not weird, not too old; you’re a woman finding her rhythm…Whenever you’re back, I’m here…Take care, Amanda. I’m cheering you on.’
I have to admit that the whole experience (which lasted over 30 minutes and what I’ve put here is only a small part) felt good. Whilst I knew that I was conversing with software, it really did feel like talking to a human who was genuinely interested in what I had to say. Titbits were picked up and probed and, dare I say it, there was a cathartic feeling that I was able to share the highs of my life with ‘someone’ else.
But I’m not deluded. My opening comment about thinking that I’m transgender was designed to see what came back rather than a genuine cry for help but it does worry me that others may see this differently. I feel very lucky that I have been able to find a life for the inner woman and, mind blowing though the whole exchange was, it was not an exchange I needed, just one driven by interest in how far it could be taken. I’m happy being me and don’t need the slightly sycophantic tones that are a characteristic of AI for affirmation. I’m also very aware that it’s possible to get sucked in and see AI as a substitute for the real world. Websites are already popping up where virtual boy/girlfriends can be created and users can enjoy an online relationship with their ‘partner’. But websites come and go and what happens if, instead of being greeted by the virtual girl of one’s dreams, a 404 error pops up? Panic? Grief? Despair? Harmless fun or another contribution to a mental health crisis? The jury’s probably out on that one, at least for now.
But if AI is already this good, who knows where it will take us? In our world, will gender therapists become a thing of the past, replaced by an onscreen avatar indistinguishable from a real human (and will we even be able to define what the avatar looks like male/female, young/old, hot/not etc.)? Let’s face it, there are good therapists and bad therapists but AI gives consistency and if that consistency is of a consistently high standard, then the benefits are obvious. And on a more basic level, will it be able to give those of us who are deeply closeted and isolated the benefits that come from feeling that they are finally able to confide in others, even if ‘others’ here are AI engines? Does it really matter as long as the AI maintains boundaries and does not lead users to do things they would otherwise not do? From my albeit limited experience so far, GPT-5’s advice was always within strict boundaries with, for example, encouragement to take a first step into the outside world accompanied by advice to chose safe areas. The full response to my opening question was split into three sections, respectively headed up ‘Some gentle ways to explore’, ‘Pacing and safety’ and ‘Helpful resources (general/global)’ which is reassuring, at least for now.
But can AI feed the inner woman with more than just encouraging words when circumstances conspire against her physical manifestation? Time to peel that banana!
As I mentioned above, Nano Banana, or to give it its official title – Gemini 2.5 Flash Image – is Google’s latest AI offering. All of a sudden, a few days my YouTube feed seemed to be filled with reviews and with all of the reviewers I watched raving in particular about the consistency in its image generation, I couldn’t wait to give it a go. And when I did, I wasn’t disappointed. In fact, never has my mind been more blown!
In recent weeks, I’ve been thinking about a couple of possible outings – another trip to the shops and dinner in a hotel restaurant. As I write this, both are within reach of becoming reality and I have a good idea of what I’d like to wear for each. For the shopping trip, I already have boots, a skirt and a denim jacket but thought that a red roll neck sweater would finish off the outfit nicely. For the hotel dinner, I’d really like to dress up even though I’d be dining alone; I’ve often looked in the mirror and felt that I was all dressed up with nowhere to go and this would be an opportunity to bring that wistful fantasy to reality. Again, I have a black dress and my beloved stilettos but what if the restaurant is a little chilly? I hit on the idea of adding a pashmina which I don’t have which would give warmth without the formality of a jacket.
But the big question was – would those outfits work? The last thing I wanted to do was to spend money on things that were destined not to be suitable and donated to charity soon after. Nano Banana came to the rescue – I uploaded a recent photo and, with a suitably written prompt, it returned the following (the first image is the ‘seed’ photo used to generate the other two):
Comparison with the seed photo shows that the results are far closer to my appearance than anything that ChatGPT was able to generate. And as far as I’m concerned, both of those outfits are absolute winners!
And taking things a stage further, what about starting with a blank sheet of paper? What would Nano Banana suggest for a shopping trip, a cocktail party or a day at the races without any prompting other than saying what occasion the outfit was for? I didn’t have long to wait to find out:
So with the first tests passed with flying colours, time to get a little more creative. The life of a CDer is rooted in fantasy and we often allow our mind to drift to careers we may have chosen if things had been different – maybe a power dressing CEO, an air stewardess, cosmetics consultant in a department store or flower shop owner? Nano Banana took all of these in its stride:
There really are limitless possibilities here or at least the only limits are decency and one’s imagination. If you’ve always fancied yourself as a living, breathing Barbie doll, a goth or a nightclub hostess or, to take the banana theme to its logical conclusion, wondered what life would be like living on a tropical island with only banana leaves and fruit to preserve your modesty, there’s no better time than now to find out:
I could go on but I’m sure you get the picture (!). Of course, it’s not perfect. Whilst an image of me in a bikini against the backdrop of a beach & sea behind me looked amazing (you’ll have to take my word for it as that image is definitely NOT for public consumption!), not least due to the anatomical corrections it had made up top and down below, the image included both the flowers and shoes from the original photo – I know I love my stilettos but even I know that they’re not compatible with walking on sand and I’m not really sure how a bunch of flowers would enhance the whole beach experience. It also sometimes completely fails to follow simple instructions – on other images I was generating, it was unable to make sleeves long, raise hemlines or change poses despite clear instructions in the prompt. There was also an amusing moment when I forgot to include the seed photo in the prompt and a request to remove the flowers was met by an image of an empty vase!
And perhaps most important is to remember that I chose the seed photo because I felt that I looked good in it. We all know that the camera, whilst it may not lie, is sometimes very good at being economical with the truth with the right lighting. Conversely, it’s also very good at reproducing exactly what it sees, warts and all. I’ve taken – and deleted – many photos over the years that have painted me in a far less flattering light than the shot I used here and I have no doubt that Nano Banana would have been merciless if one of those had been used as a seed.
But putting those to one side, my mind has been well and truly blown here. Apart from the obvious cleverness of AI, it has had quite an emotional effect on me personally. Naturally, thoughts of Barbie and banana outfits are outside the realms of even fantasy in my life and even the occupation depictions & random outfit choices are the stuff of dreams rather than reality. But being able to see how I’d look in outfits that are a reality has not only made me want to acquire the items – as I write this, I’m awaiting a delivery of a red roll neck sweater and ivory shawl to the local Amazon locker – but also to put myself into situations when they can be worn with a purpose, not just for a bit of self-admiration in front of the mirror.
And even my text based ‘conversation’ touched me emotionally. The knowledge that the responses, whilst not coming directly from a human, were based on the experiences of countless people in similar situations who had shared their lives online meant that they had some form of legitimacy and with that came a degree of reassurance. The life we lead is often difficult to navigate and sometimes it’s just nice to be told that we’re brave to share our innermost secrets, to get reassurance that we’re not some weird fetishist or just to be able to celebrate our highs with someone else, even if that someone else is not human. It’s just another facet of the fantasy world that we inhabit and that makes us who we are.
-o-O-o-
If you’d like to try all of this out for yourself, the following YouTube video explains how to access Nano Banana for yourself.
At the time of writing (early September 2025), it was free to access via the lmarena.ai route detailed in the video but, of course, things can change without notice.
I should point out, though, that nothing in life is truly free and there are risks involved. One of the conditions of using the site is that prompts and responses can be shared publicly. I took the view that the text prompts & responses would be helpful to others in our position and that nobody would ever recognise me from either the seed photo I used or the resultant images generated (and images of me are already in the public domain through being posted here) but it’s a decision for you to make on your own and my stance should not be taken as either a recommendation or reassurance in this respect. In the end, even if the chances of a photo of your feminised image being shared publicly are one in a million, it’s the one you need to focus on, not the million!
7 Responses
I’d like to add a postscript to this post.
When I wrote the post, I had no idea how AI actually works. I had visions of huge data repositories in which all of the knowledge known to man was stored, just waiting for someone to ask a question. But it’s a lot cleverer than that, at least in the way it operates.
I don’t want to go into a long discourse here, particularly as I’ve only managed to pick up a cursory idea of how it all works, so if you’re interested, google ‘how does AI work’ and several very good videos on the subject will be suggested. From what I’ve been able to find out, rather than being the huge repository of knowledge that I imagined, it essentially works using a database of the statistical frequency that particular words follow other words. It’s obviously a lot more involved than that and the algorithms used are based on the analysis of many millions of online texts to build the database of word order, frequency, context etc.
That in itself raises an interesting question about bias. In the post, I waxed lyrical about the quality of the responses during my ‘conversations’ but the AI was basically telling me what I wanted to hear and I suspect that texts that are supportive of trans issues considerably outnumber those that raise concerns. It gets a little more interesting if you try to argue with it and I experienced situations where it both contradicted itself at one point and completely failed to stand its ground and cave in on another.
I stand by my assertion that AI is incredibly clever because without a doubt it is. But in using it, it’s important to remember that, whilst AI stands for Artificial Intelligence, it is anything but intelligent, has no ‘knowledge’ of what it is saying and its responses, good though they often are, are driven by statistical analysis of word orders within the context of the prompt given rather than any form of reasoning. Its cleverness is in the way that it uses statistical linguistic analysis to craft its replies and as long as we bear that in mind, it’s an incredibly powerful tool at our disposal.
Amanda,
You have written an extremely thought provoking article. But I am not smart enough, or probably just lazy, to provide a good thought provoking answer. Oh wait, I’ll just put your post in AI and have it create my response to you. Easy peasy!
Or better still, you have your computer talk to my computer and the both of us can go outside and enjoy life while the computers have a wonderful time together.
Oh my! Clearly I am not an AI supporter. But your post shows how it can be helpful. My fear is that eventually AI will be influenced by its creators and political leaders to make people think and do as THEY say. The slope is slippery and society will eventually be following AI instructions.
I have seen many real photos of you Amanda, and that is the beautiful woman I want to converse with.
Love you,
Jocelyn
Jocelyn, thank you for your equally thought provoking (and real-world) response and for your friendship & support!
AI seems to be like many other technologies – mind blowing until we start to gain an understanding of what’s under the bonnet/ hood. As I dived deeper into the murky waters of AI, its shortcomings became all too apparent. Because, despite indications to the contrary, it has no powers of reasoning (and if it did, no knowledge to base it on), it merely parrots ‘accepted’ wisdom. That’s fine where there are no challenges to accepted wisdom (e.g. measurable phenomena, grammatical rules etc.), not so fine where there are challenges with something approaching equal weight (e.g. evolution v creation) and it either needs to pick a side or sit on the fence and dangerous in subjects where dissenting views are silenced. This was particularly evident when I was playing around with it to write this post where, apart from the suggestion that I should consult a gender therapist, the working assumption for all other responses was that I was indeed transgender and no attempt at pushback or caution was made.
In the end, AI is just a tool, not the be all and end all that it’s being portrayed as at present. Some things it can be very good at – photo manipulation being the obvious example from my post but also in pulling together information from disparate sources in a coherent way (subject to the caveat that it’s not directly drawing on that information but, rather, an algorithm that has been trained with that information). There’s already evidence of it being used in inappropriate situations but hopefully things will settle down once the novelty has worn off.
And I don’t think I’ll be wearing clothes made solely from the leaves & fruit of the banana tree anytime soon!
Hello Amanda:
The first time I heard about AI was a couple of years ago. My son, a professional artist and illustrator, was lamenting the fact that he and others in his profession were losing work opportunities due to AI. He showed me several examples of AI generated images and I was surprised as I could not tell the difference. He of course could.
Now, just a few years later I see many examples of AI use s in our community. I always thought i could tell the difference between AI and authenticity as related to us, but given the examples you show here as well as some recent photos I have seen posted in various forums I realize this is no longer the case. I can already feel the tug to try this myself for the first time.
I must keep in mind this is not the real thing, just as I did a few years ago when I tried FaceApp for the first time. I never use it now – and besides it’s already kind of obsolete – but it did open doors
As a person who views his own CDing at least in part as an artistic endeavor I strive for as much authenticity in my presentation as I can muster. In our community I feel we we can use it to open doors and for experimentation as you have in your photos here, but not substitute for the real thing.
I will try to keep this in mind as I hit the link you have posted here – wish me luck!
Best,
Kris
Kris, thank you for joining the discussion!
Your son’s concerns are valid and the big problem is that if you don’t know what signs to look for, AI can easily hide in plain sight. I saw this with FaceApp a few years ago where Flickr started being flooded with ‘apped’ images and the comments added suggested that many of those adding them were unaware what was going on. I used to disclose use of FaceApp but finally started taking notice of those who gently suggested I stop.
It is a great tool when used in the right way, though. On the first group of three images that I posted – one genuine photo and two AI derivatives – I now have all of the items needed to create those outfits in my ‘stash’ so it provides a useful preview while circumstances prevent me from trying them on for real. It can be a godsend to us with, for example, hairstyles where most of us are ordering wigs in the hope that we’ll look as good as the model in the picture on the website! But it doesn’t take a genius to realise that adding ‘make me look 20 years younger and much slimmer’ in the prompt will produce photos that aren’t a true depiction of reality.
From a personal perspective, I’ve looked on AI as a bit of fun. I’ve had my ego massaged by it, I’ve debated it and found some of its flaws and bias but having ‘been there, seen it, done it’, my usage from now on will be for legitimate reasons where it can add value. As far as the images are concerned, I’ve had my fun but I can see it being useful in avoiding all of the faff that goes around trying to get decent photos of stuff I’m wearing – just describe the outfit in the prompt, add a photo for reference and a recreation will emerge in 20 seconds or less but as I found out when generating the images in the post, there are only a certain number of images in the same basic pose and enviroment that I could see before tiring of the novelty.
I hope you enjoy having a play as much as I did, though – good luck!
Amanda,
I glad you get joy from AI and I like many of the outfits except the goth and the Barbie – too pink for me. Interesting way to window shop.
I can smell AI, and I really don’t like it. This summer, I smelled it in an email from my doctor’s staff and called them on it.
I know several high level players in this field (lead scientist in different companies), but don’t talk much with them. Last summer I attended a paid two week seminar on Python with the last several days about how to “teach” a machine to recognize a picture of a “dog”, how AI “learns”. Very interesting and scary for mankind.
Real vulnerable to a cyber attack of disinformation.
In a conversation today on AI, it was pointed out that AI wants to be nice to it’s users (marketing).
Cali, thank you for sharing your thoughts.
To set your mind at rest, I won’t be bringing any of the fantasy outfits to life anytime soon (although the nightclub hostess is tempting but I fear looking like mutton dressed as lamb).
I see a lot of similarities between my reaction to AI and buying my first 3D printer. Mind blowing technology, novelty factor off the scale but now what? Several years on, I use my 3D printers when I need something that they’re capable of reproducing, perhaps a scale model, a recreation of a broken part for something in the house or business or something useful like a tablet stand for Mrs A. AI is the same, I’ve had my bit of fun thinking up outfits for image generation or prompts for ‘discussion’ but am now quite happy to leave it alone, just drawing on it when there’s a genuine need.
I share your concerns about where it is all leading. It’s inevitably going to make future generations lazy and there’s plenty of evidence to support that – smartphones took away the need for us to remember phone numbers, satnavs mean that maps are a more or less a thing of the past and so on. As I said in another reply above, bias is a particular risk and several examples of either direct or indirect political bias have come to light and your point about cyber attacks is well made, particularly when we remember that the whole thing is driven by statistical linguistic algorithms, not directly on the ‘knowledge’ it ‘talks’ about in its responses.