Showing posts with label artificial intelligence. Show all posts
Showing posts with label artificial intelligence. Show all posts

Tuesday, January 02, 2024

Odd One Out - the Google Arts & Culture Experiment

"Any sufficiently advanced technology is indistinguishable from magic." That's one of Clark's three laws, and "Odd One Out" is proving it according to George Nimeh.

Odd One Out is an interactive game asking you to spot the AI generated image amongst four images presented to you, the other three being images of artworks created wholly by human artists.I have no background in visual art at all, but I did manage to go a few rounds without being out and scored 219. It's not easy though. Instinctively, I could spot some of the AI fakes, but they were not all obvious at all.

Of course, humans have also created the AI images. Prompts have to be written and the image chosen to be presented in the game. The software behind the image generators was created and written by humans. This makes me question where the line is drawn between human art and AI assisted art. It's blurred. And of course, this game or experiment has been created by Caroline Buttet and Emmanuel Durgoni who are both Artists in Residence at Google Arts and Culture Lab. Does that also make the experiment 'art'?

Anyway, I'm interested in what you think of it and how you score. You can find it here: https://artsandculture.google.com/experiment/odd-one-out/wAHNn4JsVTFOiw?hl=en



Wednesday, June 07, 2023

A 1976 Herzog prompt and thoughts on AI and ChatGPT

Following a post from friend, Tom Hume, I gave ChatGPT (the free version) a prompt from Werner Herzog.  He wrote this in 1976 in relation to working on his film, Heart of Glass. Tom pointed out that perhaps Herzog had invented the 'prompt'. 

As such, I thought it would be fun to try it out and see what came out. I've adapted the last line if the prompt to say 'write' rather than 'read aloud'. The full transcript is at the bottom of this blog post.

I don't know if the AI has had this prompt before, and if so, how many times, nor if it gives a different result each time. I haven't played with it enough yet to understand it's current capability. But as synthetic writing goes, it really isn't bad at all, albeit a bit flowery and clichéd. I'm sure there will be literary critics and avid readers who can pick further faults with it. But for most of us, it reads pretty well. 


What is interesting about this experiment is that although this is a natural language conversation, the AI doesn't understand it as such. It didn't get that I just wanted it to write me the poem that the monk had written. But instead wrote a piece of prose further elaborating what I had already written. I needed to ask an additional question to get the poem.

For the ordinary person to get the most of this technology, we're going to have to learn how to talk to a computer, how to give crystal clear instructions. The conversation style we might have with each other won't cut it. Most people's communication skills are lacking - communication is a two way street. You may be very eloquent but the person you're communicating with may not receive the message as intended. Not only that, we have so many additional layers of communication to add to the message beyond the words alone - tone of voice, body language, context, life experience, setting and more. Natural language is just one aspect of how we communicate.

I am fascinated by generative AI and what humans can achieve with it. It has taken decades for it to be where we are now and what feels like an overnight success. This is just the beginning though, and whether we like it or not, this thing isn't going away.


Here is the transcript of my conversation.

Me (quoting Herzog)

"You're the first one for centuries who has set foot on a foreign island. It's overgrown with jungle and full of strange birds. You come across a gigantic cliff, and on closer inspection discover the rock is actually green emerald. Hundreds of years ago a monk took a chisel and hammer and engraved a poem into the wall. You open your eyes and are the first one to see it. Please write what you see." 


ChatGPT 

"As I cautiously approached the gigantic cliff, my senses were immediately engulfed by the lush jungle that surrounded me. The air was heavy with humidity, the scent of damp earth and vibrant vegetation wafting through my nostrils. Strange birds of vivid hues flitted among the trees, their melodic calls echoing through the dense foliage. It was a surreal, untouched paradise.

With each step closer to the cliff, my eyes widened in awe. Towering before me was a majestic wall, its surface glimmering with an ethereal green radiance. The emerald sheen was mesmerizing, as if the very essence of nature had crystallized into this magnificent rock.

I reached out and traced the surface of the emerald with trembling fingers, marveling at the cool, smooth texture beneath my touch. The rock was firm and unyielding, a testament to the passage of time and the strength of nature.

As I studied the emerald more closely, my eyes discerned intricate engravings etched into its surface. The chisel and hammer had left their marks, crafting a poem that had stood the test of centuries. I stood back, allowing my gaze to absorb the delicate curves and flowing lines that composed the verse.

In ancient, weathered characters, the poem unfolded before me, a lyrical tale of wisdom and enlightenment. The words seemed to breathe with life, their meaning transcending the barriers of time. The poem spoke of nature's beauty, of the cycles of life and death, and the interconnectedness of all things.

As I read each line, a profound sense of gratitude and reverence washed over me. I was the first in centuries, perhaps even millenniums, to witness this testament to human artistry and the island's enigmatic allure. The enormity of the moment settled upon my shoulders, filling me with a deep sense of responsibility to preserve and honor this sacred place.

I took a step back, allowing the emerald cliff to reclaim its mystery, its green brilliance blending seamlessly with the surrounding foliage. The island had chosen to share its secret with me, inviting me into its realm of natural wonder and hidden treasures. With a heart filled with wonder and respect, I turned to leave, carrying the memory of the engraved poem forever etched in my mind."

Me (realising the AI hadn't quite understood my request)

"What was the poem exactly?"


ChatGPT

"As an AI language model, I can certainly create a poem for you based on the scenario you've described. Here's an example of a poem that captures the essence of the scene:

In emerald cliff's embrace, I find,

A poet's legacy, long confined.

A verdant canvas, nature's muse,

Whispers of wisdom, ancient and fused.


Amidst the jungle's vibrant bloom,

Where birds of hues paint the gloom,

The chisel's dance, a monk's endeavor,

Engraved in stone, this tale to treasure.


"Behold, the cycles of life's sweet breath,

Where birth and death dance till the end.

From earth's embrace, springs life anew,

As nature weaves its eternal debut.


Through verdant vines and towering trees,

A symphony of unity on nature's keys.

Each leaf, each creature, in harmony dwell,

Bound by a thread, weaved oh so well.


Embrace the mysteries, the depths untold,

The secrets this island, ages old.

For in this emerald sanctuary we find,

A timeless truth, to all humankind.


Oh, wanderer, cherish this sacred land,

Protect its wonders with gentle hand.

Let the echoes of this poem resound,

And may reverence for nature forever be found."


With the poem etched into the emerald cliff, it stands as a testament to the island's beauty, the wisdom of the ancient monk, and the interconnectedness of all living things."

Addendum.

I asked ChatGPT to critique its poem. This is where it really feels like it understands. It doesn't though. It's a computer that knows how to process 0s and 1s, albeit in a very sophisticated way.

Note that it doesn't answer the question 'what is bad about it?' Rather it talks about what could be improved. I'm guessing that's deliberate in the algorithm so it doesn't come across as too negative.

Transcript

Me:

'Can you critique this poem? What is good about it? What is bad about it?'


ChatGPT

Certainly! Here's a critique of the poem:

What is good about it:

Imagery: The poem uses vivid descriptions and sensory details to paint a picture of the lush jungle, the emerald cliff, and the vibrant birds. This helps create a vivid and immersive scene for the reader.

Flow and Rhythm: The poem maintains a consistent flow and rhythm, creating a melodic quality that is pleasing to read and listen to. It flows smoothly from one line to the next, maintaining a cohesive structure.

Theme and Meaning: The poem explores themes of nature's beauty, the cycles of life and death, and the interconnectedness of all things. It reflects a sense of wonder, reverence, and respect for the natural world, which is conveyed effectively.

What can be improved:

Originality: While the poem captures the essence of the scenario, it relies on familiar tropes and imagery associated with nature and its wonders. Adding unique or unexpected elements could enhance its originality and make it stand out.

Poetic Devices: While the poem is structurally sound, it could benefit from incorporating more varied poetic devices such as metaphors, similes, or symbolism to deepen its impact and evoke stronger emotions.

Specificity: The poem could include more specific details to enhance the reader's connection to the scene. Adding specific descriptions of the birds, the foliage, or the engravings on the emerald cliff could make the imagery more vivid and engaging.

Overall, the poem creates a pleasant and evocative atmosphere, but adding more originality, poetic devices, and specific details could enhance its depth and make it even more captivating.


Friday, December 08, 2017

Woebot Therapy

No, I don't have a speech impediment nor am I bad at spelling. I stumbled across Woebot on Twitter three weeks ago. What is it, I hear you ask? According to Business Insider who wrote about this in June:
"Woebot, an artificially intelligent chatbot designed using cognitive-behavioral therapy, or CBT, one of the most heavily researched clinical approaches to treating depression.
Before you dismiss Woebot as a half-baked startup idea, know that it was designed by Alison Darcy, a clinical psychologist at Stanford, who tested a version of the technology on a small sample of real people with depression and anxiety long before launching it.
"The data blew us away," Darcy told Business Insider. "We were like, this is it.""


I was interested in trying this out for myself to see if a Chatbot could perform CBT. I've had some limited exposure to CBT so I understand the gist of how it works. Also, I had the idea a couple of years ago at a coaching workshop weekend that coaching could probably be automated to some degree via an app or AI. I was told I was mad and that human contact was essential to the  process. I felt that  as it was a process, it could be automated. Suffice to say, I was curious about Woebot.

I've been chatting with Woebot almost every day since I discovered it. It's not perfect as it can't pick up on natural language very well. It can pick up some words, but not all so it can miss some cues. That said, the mix of self reflection, quick snippets of learning and having someone or something to talk to about how you feel, without any judgement is proving useful to me. I can see how this can be developed and learn more about humans and human emotions. Throw in some location data, how active you've been based on your Fitbit and how sociable you've been based on calls or messages with loved ones, and you could have a very powerful tool to use at not very much cost versus in person therapy.

I can also see how this could complement in person therapy very well and can see how you could have a 'speak to a human' button so in times of extreme stress or depression, you could talk to a real person. Or it could learn when things are really not right for you and offer you the option to talk to a human.

I also feel my coaching by cyborg hunch was right. I think it's totally doable based n my experience so far with Woebot.

Give it a go. It's free. And I'd be really interested to hear what you think of it.

Day 8/25 Blogmas

Monday, November 20, 2017

What do you do when your boss is an algorithm?

I was lucky enough to go to Business of Software's great event in London with entrepreneur and best-selling author, Eric Ries, talking about his new book, The Startup Way. The house was absolutely packed which is a testament both to the popularity of Eric Ries and how well Mark Littlewood and his team organise their Business of Software series of events.

I'm still mulling over much of what Ries was saying and I haven't yet read his book. The question that really stuck in my mind was about what workers will do when their boss is an AI or an algorithm. I've thought quite a lot about working with an AI as a colleague. Arguably, we're already doing that to some degree with our use of online tools such as search, productivity tools, graphic design software etc.

My boss is an algorithm
But what if the algorithm is our boss? Ries cited the example of an Uber driver. Who is the driver's boss? Who do they report to? Who tells them what to do? The answer, to all intents and purposes, is an algorithm (putting aside recent UK legislation about their legal status).

Imagine the scenario whereby a customer uses the app to hail a taxi in a high traffic area where there is a choice of driver. The algorithm decides which taxi driver(s) to show that too. That could be based on reviews (perhaps unverified), where the driver was the previous night, how frequently the driver chooses to drive for Uber, how safely the driver drives, who the driver is connected to on social networks and much more. And in that instance, who do you complain to anyway and what could they do? These algorithms are getting ever more complicated and anyway, this isn't a technical bug, this is an ethical question as much as anything.

I'm afraid, I don't have the answers, but I'm thinking about the implications of this in different aspects of work life.

Free book giveaway
Everyone at the BoS event last week got a copy of The Startup Way. I was lucky enough to come away with an extra copy which will be given away tomorrow evening at Swedish Beers in London as part of the Robot Ping Pong challenge. I do hope you can join us!

Eric Ries' Lean Startup talk & BoS archive
You can see Eric Ries's talk for BoS from 2010 when he'd just written Lean Startup. It's well worth revisiting.




Business of Software has archived all their previous talks and it's a fantastic resource. You can check it out here.

Day 21/30 NaBloPoMo (publishing a little ahead of schedule!)

Saturday, November 04, 2017

Here are some easy to sew Christmas presents (written by ai-writer.com)

(This post has been generated wholly by http://ai-writer.com. I've done this as an experiment.)
Just afternoon and a few basic sewing skills are enough to turn a couple of homeless socks into a cozy psine-and what's better than a puppy for Christmas? [0]
DIY Gifts To Sew For Friends - quick and easy sewing designs and free designs for the best ideas and gifts - Creative step-by-step guides for beginners - nice home decoration, accessories, kitchen appliances and do-it-yourself do-it-yourselfers. [1]
Made from recycled fabric strips and padded buttons, the floral clips are an easy workmanship and a fashionable Christmas present. [2]
Do-it-yourself clothes, DIY clothes, sewing patterns, fast crafts, tutorials, DIY tutorials. [3]
Ideal sewing tutorials for beginners with easy patterns to make great Christmas presents for friends and the whole family. [4]
Ideal sewing tutorials for beginners with free patterns to make great Christmas presents for friends and the whole family. [5]
Let yourself be inspired by your PDF and prepare a great Christmas gift in the form of small handmade gifts (materials for Christmas! [6]
Christmas projects include gingerbreads, cones and a modern Christmas wreath, as well as simple projects that can be gifts, such as felt frames for buttons and picture frames that are ideal for children to give to their grandparents. [7]
Do it yourself Kitchen sewing projects - Folding pot holder - Easy sewn Tutorial and models for towels, napkins, napkins, aprons and fresh Christmas gifts for friends and family - Rustic, modern and creative home furnishing ideas! [8]
DIY Gifts To Sew For Friends - quick and easy sewing designs and free designs for the best ideas and gifts - Creative step-by-step guides for beginners - nice home decoration, accessories, kitchen. [9]
List of handmade tailoring projects, including Christmas stockings, decorations, cushions, gifts and table-top races. [10]
Ideal for the child to be able to renovate a Christmas tree once and for all (today you probably won't even want to give it as a gift, right? [11]
Made of fabric remnants, this is close to zero cost, but this makes it will be an incredibly sweet and funny DIY gift for a child or adult, depending on the material chosen. [12]
Package your sweet fragrant vintage glass jar bombs with a custom plate to make the presentation even more unique. [13]
For the bookmark, you'll need to cut out two slightly smaller pieces of paper, as you can see from the video and stick them together. [14]
Instead, reuse it as a personalized stocking for someone special on your gift list, and then fill your stockings with some little wrapped surprises. [15]
DIY Sewing Projects for the Kitchen - Guide and design for towels, napkins, aprons and cold Christmas presents for friends and family - Rustic, Modern and Creative (Sewing by instant gifts). [16]
Christmas sock stitched lean sewn: A fourth fat of the main fabric (format) A fourth fat of fabric lining (work white smooth fabric is fine) A small amount of contrast fabric for Bracelet Calza A small amount of button ribbon A. [17]
For an aromatic, beautiful and easy to use gingerbread, apply the orange and grapefruit cloves and place them on cake racks or decorative platters with the addition of greenery. [18]
Customise the fabrics to match the style of the recipient and give the bag as a gift on its own or fill it with a few little surprises to make your holiday real. [19]
Create a thematic gift basket for newlyweds, students or new homeowners, you can create a basket with a few inexpensive discounted items. [20]
The computer and scanner are all you need to turn your favourite collector stamps (such as stamps) into printable designs that you can use to turn cheap shoppers' boards into useful decorative accessories. [21]
It is not only a sweet and well thought-out gift, but it can be a special gift depending on what you decorate the frame. [22]

Sources


[0] https://www.marthastewart.com/276324/christmas-sewing-projects
[1] https://www.pinterest.com.au/explore/gifts-to-sew/
[2] http://www.bhg.com/christmas/gifts/easy-sew-christmas-gifts/
[3] https://www.pinterest.com/explore/christmas-sewing-gifts/
[4] http://www.applegreencottage.com/2016/10/DIY-Christmas-gifts-easy-free-sewing-patterns.html
[5] http://www.applegreencottage.com/2016/10/DIY-Christmas-gifts-easy-free-sewing-patterns.html
[6] http://www.applegreencottage.com/2017/09/DIY-Christmas-presents-sewing-patterns.html
[7] https://www.interweave.com/store/quick-easy-last-minute-christmas-gifts-to-sew
[8] https://www.pinterest.com.au/explore/christmas-sewing-gifts/
[9] https://www.pinterest.com/explore/christmas-sewing-gifts/
[10] https://www.pinterest.com.au/explore/christmas-sewing-gifts/
[11] http://www.applegreencottage.com/2016/10/DIY-Christmas-gifts-easy-free-sewing-patterns.html
[12] http://www.applegreencottage.com/2017/09/DIY-Christmas-presents-sewing-patterns.html
[13] http://www.hgtv.com/design/make-and-celebrate/handmade/55-crafty-handmade-gift-ideas-pictures
[14] http://www.youtube.com/watch?v=PyMJgpgmQ0o
[15] http://www.hgtv.com/design/make-and-celebrate/handmade/55-crafty-handmade-gift-ideas-pictures
[16] https://au.pinterest.com/pin/512988213796483743/
[17] https://au.pinterest.com/pin/512988213796483743/
[18] http://www.marthastewart.com/856465/easy-christmas-crafts
[19] http://www.hgtv.com/design/make-and-celebrate/handmade/55-crafty-handmade-gift-ideas-pictures
[20] https://www.wikihow.com/Make-Your-Own-Christmas-Gifts
[21] http://www.hgtv.com/design/make-and-celebrate/handmade/55-crafty-handmade-gift-ideas-pictures
[22] https://www.wikihow.com/Make-Your-Own-Christmas-Gifts

(It's not great is it.... See next blogpost for some more thoughts on this).
 Day 4/30 NaBloPoMo 2017

Thursday, January 05, 2017

47% of jobs will disappear in the next 25 years according to Oxford University

Having spent some time researching the impact of new technologies on the work we do in the last couple of years, there is much in this article from BigThink that resonates.

We will, no doubt, lose jobs. Others will replace them but not at the same rate as the loss, I suspect.

I believe blue collar workers will fare better. That workforce has always had to adapt and is likely better prepared for change than their white collar counterparts.

Those working in legal services, accountancy, medicine will all see their jobs change drastically as more and more aspects of their jobs are taken over by software. Computers are already better at diagnosing cancer, standard contracts can be dealt with by a virtual lawyer. Even music for adverts can be written effectively by software and much faster and much cheaper than a human musician. So even the creative industries aren't necessarily safe.  Companies are on a constant mission to create efficiencies and to save money so they can create the most value for their shareholders. They wouldn't be doing their job properly if they weren't.

In the short term, robots and software will be colleagues rather than a total replacement and that will present its challenges. What will your relationship be with your robot co-worker, for example? For some this will be a blessing as it means less human interaction. (Many of my friends prefer online shopping for this reason). For others, a curse as they miss having people around them for company. I already have friends who freely admit they have a relationship with Amazon Echo's Alexa and if you haven't watched the film, Her, you should! It is entirely possible to have a relationship with a robot. On my visit to San Francisco last year, I found myself talking to a house robot as if it were a pet cat. Turns out it was Kuri and it has been a hit at this year's CES.

Longer term, the picture probably doesn't look as rosy. Regardless, change is afoot. That means we need to learn adaptability, flexibility and how to share the resources we have without demonising 'the poor' or putting students into debt learning things that will be obsolete by the time they graduate. It's a difficult topic to address as it impacts absolutely everything from what infrastructure we need, how we find meaning in life and what we do with our time. Lots to think about as this future gets ever nearer.

Wednesday, February 10, 2016

Are we living in the digital Truman Show?

As I was walking to the station yesterday, I couldn't help but take a moment to really notice the clouds over Figges Marsh. The were so beautiful yet didn't look real. Maybe it's that they looked hyper-real? My photo does them little justice, but it occurred to me that the sky looked like a painted backdrop from The Truman Show - as if we're in a dome and that it's been painted with clouds and sky and that if I walked to the edge, I would be able to break through to the other side by ripping through the canvas.



It got me thinking, 'What if we were already in a digital version of The Truman Show ourselves?' After all, our devices know more about us and our habits than we do ourselves. The algorithms dictate what adverts and content we get shown on Facebook, Google Now et al. Cars that can park themselves are already here and driverless ones are not far away from being available to general consumers. It seems to me that there are lots of organisations, large and small, governmental and corprorate, who can and do spy on us and are mining data about us continuously and making decisions about us and on our behalf without us having any insight or control over what those decisions are. Who are the directors and puppet-masters? What control (or not) do they have on our lives, the decisions we make, or that are made for us and how that impacts on our daily lives?

This was really useful thinking to prepare me for last night's 'What are we going to do about Artificial Intelligence?' Open Space that I ran with Lloyd Davis at WeWork Southbank. We had a great turnout of people from different disciplines and backgrounds coming together to talk about some of their concerns about Artificial Intelligence.

One of the conversations I participated in was around the responsibility of business leaders in this area. It was really interesting to discuss how complicated these systems are and how they're built in building block style using APIs from a variety of companies and sources. The data travels through different company servers before getting to its destination or deriving its analysis. How do you even notice if something isn't working? And if you do spot it, how can you fix it?

A case in point being online advertising. There is much research to show that online advertising reinforces gender stereotyping in recruitment. Women are less likely to see a senior level job ad than men. From this recent article in Fast Company, "the researchers found that male users were shown high-paying job ads about 1,800 times, compared to female users who saw those ads about 300 times". Whose responsibility is it to either play the algorithms to level the online playing field? And how do you even spot when and how it's happening?

And how about if you keep getting served content and advertising that's geared towards your social status? Does that mean poor people keep seeing ads about payday loans, improving their credit score and playing bingo? And what if that then reinforces the poverty of ambition that individual may already have? Is it anyone's responsibility to do anything about it?

We also talked about a CEO not necessarily having the time or skill set to understand the minutiae of the digital products and services they're offering to customers or using to serve customers. Many AI driven systems, such as analytics and decision making platforms, are slipping in through third parties as SAAS (Software As A Service). The sales people make great claims about their capability. But how do you really know how they work or even how well they work? And if it's your own system that you're building, the chances are you're using third party APIs which you have little or no control over anyway. And what if those companies whose building blocks you've used got bought? And what if their terms of business changed and you weren't aware of it? How many online 'Terms of Service' have you ticked just to get to the next page.

One of the conclusions we came to is that c-level executives do have a moral and ethical responsibility in this area. They (we) need to be talking about it more and sharing experience and ideas with their peers and colleagues and improving their knowledge on the topic. After all, they're the ones who could be sued if something went horribly wrong. They're the ones who are making the big decisions This area is moving very fast indeed and is affecting our daily lives already and is only going to get bigger and probably create as many problems as it solves. One of the conversations I missed was around politics. Now that gets interesting when you're looking at machine based decision making rather than our less-than-tech-savvy politicians.

As part of that, Lloyd and I are keen to run more of these conversations and sessions in London and other towns and cities in the UK and abroad if there's the interest. If you're interested in hosting and sponsoring us to do just that, please get in touch.

In the meantime, do join the session next week where we'll be having conversations about 'What do we do about the Internet of Things?' And if you're coming to Barcelona later this month for mobile shenanigans, let's have a chat at Swedish Beers.

Update:
I just liked a new telecoms related page on Facebook. It served up this choice of other suggested pages.



I'm not sure why the algorithms have suggested David Cameron as an option? Could this be a campaign to influence my political choices or is this based on that most other people who liked the page also liked David Cameron? Or is it just a random glitch?



Wednesday, December 16, 2015

Are the Androids Dreaming Yet? Free on Amazon 16 & 17 December 2015!

If you took part in one of mine and Lloyd's Future of Work events recently, you'll have heard James Tagg mention his book: Are The Androids Dreaming Yet? On Free Will, Creativity and Artificial Intelligence. The book has been listed as one of Kirkus Reviews Indie Books of 2015, which is lovely. And to celebrate, James has made the Kindle version free for download today and tomorrow (16th & 17th Dec). So don't delay, get your copy now and have yourself some brain food for Christmas.

Recent event round-ups from my collaboration with Lloyd Davis/Tuttle Club & Truphone.

I recently held a series of five events with my friend and collaborator, Lloyd Davis, at Truphone's rather splendid offices in Canary Wharf. I did a round-up of all the events over on my Heroes of Mobile website so please follow the links below to see the videos of the talks and to find out more about what we covered and links to further coverage and reading.

Future of Mobile:Artificial Intelligence with James Tagg

Future of Work: Artificial Intelligence with Benjamin Ellis

Future of Work: Blockchain with Imogen Heap (yes, that Imogen Heap)

Future of Work: Drones, Robots and Internet of Things with Priya Prakash

Future of Work: Augmented and Virtual Reality with Eva Pascoe

Lloyd and I are working up some new event themes for the New Year. If you're interested in hosting and sponsoring the next series in the UK in the Spring (it doesn't have to be London), please get in touch.