Fortunately/Unfortunately I’ve been into NFTs since they hyped into culture late 2020 and though they have their merits, most are garbage and exist solely as speculative assets and/or outright scams.

When hype faded, controversies piled up and market value crashed, NFTs developed and cemented their highly negative reputation. This seems mostly irreversible and in a lot of cases well deserved but this is actually a progressive step as the term “NFT” is far too ambiguous.

So what is an NFT?

NFT stands for “non-fungible token” and it’s essentially just an entry in a database that says you own something with utility. Think of it as a technology.

This is most commonly understood as owning digital art. An image of a bored ape or a video of a tree growing out of Madonna. Art is the utility.

But how can you own a specific media file? Images and videos are infinitely replicable and thus essentially free. Without scarcity it’s difficult to derive value, but if you can instead create a limited amount of tokens that represent ownership of a media file, then you introduce scarcity, which results in value.

Less appreciated is that an NFT, the token, can represent “ownership” of anything. A song, membership to a club, a story, rights to a story, a share in a house, a share in a company, a promise to meet up for a pint… anything.

But how?

• A database entry states that you own a token that represents a thing with utility.

• The entry is accepted as genuine.

• The creator of the token acknowledges the above is true.

That’s it.

It’s all just people agreeing that entries in a database are what they say they are and if enough people believe in it, it exists.

This also means that if people stop believing, or collective sentiment shifts, then what the token represents changes too.

So if NFTs are essentially just a database entry, and they can represent ownership of anything, then the term as it’s currently used and understood is too ambiguous. In most contexts it should be retired.

Let’s refer to “NFTs” for what they really are:

• Art

• Architecture

• Access passes

• Proof of attendance

• Clothing

• Avatars + Identity

• Dance moves

• Weapons

• Club membership

• Land

• Seed investment in a startup

• Voting rights in an organisation

• … and so forth and so forth

… then we can discuss them with more clarity.

What is useful about this tech?

NFTs are generally stored on public blockchains (databases) and interact with smart contracts (computer programs). This means:

• Token information and ownership is transparent and publicly available

• Data cannot be corrupted

• Entries to the database are recorded in seconds

• Tokens are intrinsically linked to a digital currency

• Computer programs can be automatically executed when transactions occur

• No single entity owns or controls the database

This means that you can:

• Buy and sell tokens that can represent anything, in seconds, to anyone, with confidence.

• Prove you own or did something, or many things, in seconds, to anyone, with confidence.

• Automatically execute many actions on every transaction, in seconds, with confidence.

When these intrinsic features are applied to the many things we do today, they have the potential to greatly increase efficiency and reduce friction, thus enabling activity previously considered too costly, infeasible or impossible.

I’m sure you’ve thought of some great examples already.

Can’t this just be done with systems that already exist?

Yes but it’ll require the owners of those systems to integrate and implement these features. It’ll require deals, agreements and consensus. The effort involved is not trivial and in most cases infeasible. There is value in a tech’s inbuilt efficiencies.

Relying on people to uphold a token’s utility seems risky.

It is but isn’t pretty much everything in a society reliant on people holding up their end of the bargain? Society has laws, ethics, morals, enforcement, and NFTs are not exempt from them.

Is there anything else worth noting?

Digital things and the ownership of, have been building for decades, and the culture and behaviour born and reinforced by NFTs over these last couple of years will also influence the mainstream. Even if the tech fails completely its culture and behaviour will be its legacy.

(something I hope to explore in future newsletters)

I skipped over your scammy content. Why are you trying to scam me mister scam artist?

You are entitled to your opinion but you are mistaken. I’m not trying to scam you but there are a lot of scams and the internet can be a scammy place, so perhaps you are right to assume everything is a scam... Trust no one!

Anyhow here’s a scam-free Deca gallery of the first year of The Tickle – a zine for contemporary digital art & creative writing. It’s brilliant and features loads of artists creating work on Tezos (the art blockchain) and each issue is an NFT. A magazine as an NFT? Oh no…

It’s a wonderful world…

Links to restore/erode your faith in humanity.

• Nick Cave is definitely not a fan of ChatGPT. [link]

• We can now duplicate anyone’s voice in three seconds, hooray?! [link] [link]

• Generative AI will supercharge game development, a medium heavily reliant on asset creation (swap game with metaverse for hype). [link]

• Human movement as PII (personally identifiable information) [link]

• Create music from text because musicians have it so easy. [link]

• AI-created images lose copyright case. And so it begins… [link]

• This Rocket League bot is rad to watch. [link]

• Pet fish smarter than us knows to read the TOS.

• #livinintheusa #momlyfe #iusedtoworkingamesbutnowiprotectchildren

  

  @mobile_mamaIs the Rezvani Vengeance the safest vehicle for you and your kiddos or what, Mama? #rezvani #luxury #rezvanivengeance #cartok #momsoftiktok #foryou #foryoupage #fypシ #fyp #fy #mom #family #military

  Tiktok failed to load.
  
  Enable 3rd party cookies or use another browser

Random Idea Section

I promised an ideas section every newsletter so here’s one about AI as a pet familiar.

But first… The Disclaimer (repeated every week)

Ideas will be of dubious quality / potentially shit. Also ideas are a dime a dozen, the value is in execution, or is it?, there is no such thing as an original idea etc etc, so if you have any comments, or you’ve seen any of these ideas in reality, please let me know. I’d be very interested!

Humans love to personify things. Plants, vehicles, abodes, weapons, tools … it’s in our nature.

For now the most well-known AI personalities, Alexa and Siri, are intentionally branded, depersonalised and owned by corporations. Google is still viewed as a tool rather than an “It”. AI in sci-fi tend to be self-aware entities, fast-forwarding to the end game.

I believe there’s a huge opportunity instead for AI to sit somewhere in between as a kind of pet familiar. What’s a familiar?

In European folklore of the medieval and early modern periods, familiars (sometimes referred to as familiar spirits) were believed to be supernatural entities that would assist witches and cunning folk in their practice of magic. … they could manifest as numerous forms, usually as an animal, … and were described as "clearly defined, three-dimensional... forms, vivid with colour and animated with movement and sound", as opposed to descriptions of ghosts with their "smoky, undefined form[s]". [source]

AI will become more personalised as we begin to train them on our own data and preferences, thus separating their identity from the companies that create them. We’ll give them names and imbue them with their own personality, but most crucially, their own visual representation.

It’s this visual representation, and thus contextualisation, that’s key. AI is in need of a form. A glowing orb is a start. A paperclip with a face is better 👀📎 but one just needs to look at toys and games for precedence – dolls, soft toys, Cars, Tamagotchi. Children’s entertainment is all about non-living things that are alive. Our identities online, in threads, social media, games – our avatars – are often entirely made-up and non-human.

And frankly it’s too awkward to have AI that looks and talks like a human. It’s too close to slavery. Animal pets though are socially acceptable, and the “familiar" aspect, the granting and assistance of magical powers, is the utility that AI technology provides.

So perhaps this is how AI beings enter the mainstream. A loyal companion that one can summon like a Pokemon or Patronus, that’ll stick by your side through thick and thin, that’ll help you live your life, enhance your life, that your friends will refer to by name, as they play and socialise with their own beloved familiars.

Finally the end…

It took awhile for my second but better late than never. The next will come quicker I promise! 🤞

If you liked or hated this, please let me know. If think others would enjoy this newsletter, please share it around and get them to, as they say,
→ like and subscribe ←

I would be eternally grateful 🙏

Subscribe now

Some of you will know a little about generative AI. Some none at all. Some of you will be sick to death discussing it. This newsletter is for you all… 😅

But seriously I’m mainly writing about this topic to get some thoughts in order and also because it feels like a significant leap that’ll impact society, our livelihoods, and our sanity. A technology we’ll need to be aware of, live with, guard from, and take advantage of.

If you haven’t been fortunate/unfortunate enough to be exposed to all the brouhaha, the following is a short summary. If you know all about this already feel free to skip past this bit but you will miss out on some cute animals 🦥

What is Generative AI?

The current crop of generative AI, in general*, allows you to type something in a textbox – a prompt – which then generates a response.

The AI’s response can then take the form of an image, text, video, piece of music, your livelihood, your calm…

_

For example, I invoked the AI gods to generate images of:

A post apocalyptic London where sloths who wear cool hats have finally taken over the human race.

… and some seconds later it responds with these cute af images.

These images didn’t exist previously.
It’s not our lord Google regurgitating what it found.
It took seconds to generate.

_

Another example, I asked ChatGPT to:

Explain typical millennial social behaviour in a humourous way.

… and some seconds later it respond with this 🗡 straight to the 🖤

Well, let's see...typical millennial social behavior can be summed up as follows:

1. Spend hours scrolling through social media, occasionally stopping to "like" a post or leave a snarky comment.

2. Meet up with friends at a trendy coffee shop, preferably one with artisanal pour-over options and at least three different types of milk alternatives.

3. Take a million selfies to document every single outing, because how else will anyone know we're living our best lives?

4. Use words like "lit," "salty," and "literal" as much as possible, even if we're not entirely sure what they mean.

5. Debate the merits of various avocado toast options at brunch. (Pro tip: get the one with the poached egg. Trust me.)

6. Get outraged about everything, all the time. The world is a scary place, and it's our job to let everyone know about it.

7. Use the term "adulting" to describe even the most mundane tasks, because paying bills and doing laundry is just that hard.

8. Embrace our inner hipster by claiming to like things before they were cool. Bonus points if we can name a really obscure band that nobody's heard of.

All jokes aside, of course, not all millennials fit this stereotype, and it's important to remember that people of all ages can engage in these types of behaviors.

Once again this was entirely made up, in seconds, because we deserve the ridicule.

_

This is all possible because humanity has figured out how to train machines to recognise patterns by feeding them a bazillion zexxabits of data and content, often crawled from the internet, classified by humans, while applying algorithmic magic to guide them. These learnings can then be used to generate entirely new things quickly. Very very quickly.

So now you get the gist of it.
It’s awesome and terrifying and intriguing and x.

What’s the brouhaha?

If you use these AIs for any length of time you’ll soon realise that even though it can be extremely good, it’s often riddled with errors and inaccuracy. Hands have too many fingers, stats are inaccurate, it can’t understand morse code, basic math is bad…

It’s also really difficult to get it to generate exactly what you want from a single prompt, or even continual prompts with lengthy guidance. It feels like magic, but a little chaotic, but for many things in life 70% is good enough, and you’ll know it’ll get better fast.

Generative AI has been dominating the zeitgeist of late for a whole host of reasons but mainly around these three concerns:

1. Automating creativity will destroy jobs.

2. Feeding humans to machines (fine, human output to machines) without consent and renumeration is IP theft.

3. Generated output contains errors and inaccuracies confidently expressed.

These are all valid and real concerns and in general I agree, but also it cannot be stopped.

“Can you elaborate on these concerns?”
I probably should in the future. This is more a short summary and a focus on generative AI’s utility.

“Should it be regulated?”
We should try.

“How should it be regulated?”
This was intended as a short summary, sorry.

“So if we can’t stop it, what should we do?”

Generative AI should be considered as and utilised as tools, that grant humans what currently feel like superpowers. And like any tool, it’ll make certain tasks more efficient, it’ll inspire, and it’ll be controlled by humans.

1. The human creates and inputs into the AI tool

2. The tool generates output

3. The human curates and finesses output

4. The human re-inputs the finessed output

5. Repeat steps 2-4 as necessary

6. The human is satisfied and progresses to the next step of the process

This is exactly like any other tool. Sure it may feel like generative AI is doing a lot more than other tools. It is. But the tool is only a part of a complex process that is always sandwiched by humans, within a wider system that always involves humans. And since it begins and ends with us, the time and effort saved by these tools can be refocused on what we do best.

_

Examples please?

It’s crucial to keep in mind generated output is good but imperfect. The focus should be on tasks that don’t require a high level of accuracy (no taxes) and/or tasks that benefit from novelty and fuzziness (creativity, art), as you will need to curate and review the output. Save time, do more.

Let’s get the most obvious use cases out of the way.

All generators can be used for brainstorming and communicating ideas. Anyone can find this useful. Literally just start your prompt with: “Give me some ideas for …”

Image generators like DALL·E 2, Midjourney, and Stable Diffusion can be used to create good enough images of whatever you want. Need stock images for a brochure site, news article, pitch deck, your newsletter that nobody reads 🥲, art for nfts? Aces.

Text generators like ChatGPT and GitHub Copilot can be used to write pretty much anything you want. Essays, marketing copy, functional code, stories, poems, speeches. It’s quite convincing with conversations so could be used for certain kinds of customer service and support. It can be more useful than Google if you just want to get a summary of something.

Finally it’s worth mentioning that some people note that generated output can be nice to look at, well-formed and even accurate – but boring, repetitive, and of a discernible AI style. Once again that’s where the human comes in. You started prompting it. Now do what you do best and build upon it. Take over.

The wider world has only just started thinking and experimenting with how best to make use of these generative AI tools. GPT-3 (the AI model that kicked off this wave) was only released to the general public barely a year ago, but already there are all kinds of interesting use cases. The future looks bright!?

_

Fine… got links?

Here are a few:

• AI as a creative partner. [link]

• Prototype of the future of Search? [link]

• Illustrated stories with a prompt. [link]

• Ads in seconds. [link] [link]

• Replace objects. [link] [link]

• Suggest code in your editor. [link]

• AI images without using AI. [link]

• Physical toy to animated controllable AR. [link] [link]

• Physical toy to marketing image. [link]

• Talk to your inner child. [link]

• Directory of AI tools. [link]

I’ll be sharing more generative AI related links in future newsletters.

I need a break from this dystopia you’re dressing up as a utopia. Show me some pretty things!

Did you say you wanted more sloths?

_

So some of you may be aware I’m really into NFTs…

Oh… you’re still here? Well in that case, I have a lot of things to say and most of them are negative but there are many aspects I feel are intriguing and 😱 positive.

You will hear more about web3 from me but for now please check out this gallery showcasing the AI GATEKEEPER series by Mihai Grecu

Annoyingly the gallery is best viewed on desktop as autoplaying videos aren’t a reliable thing on mobile.

Disclaimer: Am an avid collector of their work.

Random Idea Section

So I believe I promised an ideas section every newsletter so here’s one I came up with the other day while chatting to a friend about generative AI.

But first… The Disclaimer (repeated every week)

Ideas will be of dubious quality / potentially shit. Also ideas are a dime a dozen, the value is in execution, or is it?, there is no such thing as an original idea etc etc, so if you have any comments, or you’ve seen any of these ideas in reality, please let me know. I’d be very interested!

So anyhow, wouldn’t it be neat if a fashion brand offered a tool where you could upload photos of yourself so that their AI could be trained to generate images of you styled in their latest collection?

Currently it’d be challenging to create super realistic error free images of you in their clothes but this could be ameliorated by applying an aesthetic style with an inherent tolerance for error, like being painted, or illustrated, or anime’d, or x. In fact the error prone nature of the AI image could itself be a style.

Fashion is already highly inspirational and aspirational so why not dial it up and sideways a few notches and go beyond just looking at unrelatable models or trying on clothes and looking in a mirror. At the very least you’d get cool pics to post all over social.

Thinking slightly harder, this seems like it could actually be technically feasible. Train an AI model with the latest collection. Train the model with their likeness by getting the user to upload photos of themselves, potentially faces could be enough. It’ll take a little bit of time and probably cost a bit for compute (training and generation) but it seems doable. The trick is of course balancing speed via automation with aesthetic quality. There could be potentially legal or cost issues.

Anyhow would love to know how far one could get with this. If you see anything along these lines please let know!

Finally the end…

That’s it for the first real issue of this newsletter. Hope it was somewhat interesting and enjoyable. All comments and feedback are of course very welcome!

If you think others would enjoy this newsletter please share it around and get them to, as they say, → like and subscribe ←

I would be eternally grateful 🙏

Subscribe now

* A prompt can also be combined with other inputs (e.g. images) and you don’t need to literally type in a text box, but text prompts feel more magical and everyone understands how to Google.

So what’s with the name? A long time ago I purchased the domain name atpoi.com after yet another nascent idea, never executed.

It stands for “At point of i…” which can then be completed with the word inspiration, or intersection, or incompetence – you get the idea.

Anyhow those of you that know me know I tend to share thoughts, opinions and links on things I find interesting and amusing in regards to technology, art and creativity. Yes you may now 🙄 but I will try my best not to bore you.

So this newsletter will be stuff like that but in your inbox so that it takes less effort, which means there’s a slightly higher chance that you’ll read it.

I’ll endeavour to introduce and explain newer things as simply as possible because if there’s one thing I’ve learnt in life, is that we all don’t know shit and assuming everyone has the same base knowledge is naive.

I’ll summarise what I come across so you don’t have to wade through the sludge yourself. Sometimes I’ll blurt out ideas because they seemed interesting at the time. Also I’ll be inflicting upon you all random experiments which you may enjoy, or not, who knows!

If you’re still interested, please subscribe and spread the word to whoever you wish to inflict this newsletter upon. I will be eternally grateful 🙏

Subscribe now