AI is clumsy, disrespectful and biased (again), Vitalik Buterin on biometric proof of personhood, and giving products a digital passport

Hi everyone, thanks for coming back to Customer Futures.?

Each week I unpack the ever-increasing and disruptive shifts around digital wallets, Personal AI and digital customer relationships.

This is the weekly STORIES edition. If you’re reading this and haven’t yet signed up, why not join the growing community of senior executives, entrepreneurs, designers, regulators and other digital leaders by clicking below.?

To the regular subscribers, thank you. ??

Subscribe now

?? STORIES THIS WEEK

When the motor was invented, no one could have predicted that it would end up in kitchen fridges dispensing fresh ice. Or in car door wing mirrors.

With innovation, only one of two things can ever be predicted:

• We can know?when?something will happen, but we can’t know precisely?what?it will be, or the impacts.?Like the introduction of a new regulation: We might know the date something will become law, but we can’t see how the market will respond, or the unintended consequences (GDPR => new consent requirements => consent checkboxes everywhere => fatigue and tick box blindness => online dark patterns by default)
• We can know?what?will happen, but can’t know precisely?when. Like the adoption of smartphones and electric vehicles. For decades you could have predicted that compute power and battery technology would both improve sufficiently and shrink enough that we could turn mobile phones into small computers, and produce digital-first vehicles that can drive themselves. But we didn’t know how long that would take, or what adoption would look like.

With digital wallets and digital ID, it’s the latter.?

Why?

Look at Apple Wallet, and specifically Apple Pay as an example. Adoption has taken much longer than anyone - including Apple - could have predicted. Tim Cook launched Apple Pay in 2014, but by 2016 only 10% of Apple devices even had Apple Pay activated.?

But it grew slowly. Just before Covid, Apple Pay activation reached 50%. By the end of 2022 it was over 75%, and climbing. It’s growing closer to ubiquity every day. But it’s taken over eight years.

Because adoption of consumer tech happens in two ways.?Gradually, then suddenly.

In 2014, only 3% of US retailers accepted Apple Pay. It’s now closer to 90%.

Back to digital wallets and identity. We know?what?will happen. We just don’t know when. Or how long adoption might take.?

We know that data processing is moving to the edge, onto our mobiles. We know that people will be able to run their own Personal AI applications using their data. We know that powerful new ‘verifiable credentials’ will become a standard way to help data flow between people and across businesses. And we know that digital wallets will be embedded into our everyday tools so we know who is who, and what is what.

But we also know there’s a half-life to consumer technology development and adoption.?

It’s getting faster and faster. You only need to look at the explosive growth of ChatGPT to see what’s possible, when a modern digital-first company finds product market fit.

And the market for digital wallets and Personal AI is heating up.?

More and more announcements every day. More and more startups coming out of stealth around portable personal data ‘control’ and ‘ownership’. More and more use cases for AI to help people in their daily lives, to solve real-world pains.

Put these things together and we might just be on the cusp of some pretty intense exponentials. Because innovation is so often about?combining?two existing ideas together in a new way.

Digital wallets + Personal AI

Portable digital identity + smart agents

No-code digital assistants + decentralised data storage

Customer innovation ‘at the edge’ is about to become a Big Deal. I believe it’s now clear that the breakthrough technologies listed above are now a certainty. We just don’t know a)?how?they might combine, or b)?when?they’ll hit the jackpot with product market fit.?

When they do, fasten your seatbelts.

My Personal AI will be able to spin up a new personal data application on the fly, to meet a particular need or transaction. And it will be fed all the verifiable data it needs to get something done on my behalf. All auditable. All verifiable. And all permissioned.

And then it will be deleted. Until the next time something is needed, and the application will self-assemble again.

I’ve seen real demos of this stuff in action. It’s coming.

Yes, it’s early innings in this new market. We’re nearly at the point where Apple Pay was in 2014. And perhaps where the browser was in 1994.

But there’s a difference today, post Covid. Things are remote first. Digital first. Mobile first. The adoption cycle has changed.?

And with AI, the innovation cycle just changed too. Things that used to take decades now can happen in months. And things that used to take years could happen in weeks.

I’m excited. It’s all to play for.?

When you zoom out you can see the ‘customer stack’ pieces coming together. New capabilities that will fuse into a new and explosive market?on the customer side.?

So we can now see the early parts of the customer?‘what’.

As for the customer?‘when’? Well as they say:?the best way to predict the future is to create it.

Welcome to the future of being a digital customer. And welcome back to the Customer Futures newsletter.

In this week’s edition:

• Digital Wallet UX - what is the state of the art?
• The Power Of DIDs - giving products their own Digital Product Passport
• Apple patents AirPods to measure bio-signals and brain activity?
• AI is clumsy, disrespectful and biased - again
• Vitalik Buterin weighs in on biometric proof of personhood
• 'FraudGPT' is tailored for sophisticated attacks
• International trade goes digital - using a smart identifier
• … plus other links about the future of digital customers you don’t want to miss

Let’s go.

Digital Wallet UX - what is the state of the art?

The market for portable digital identity is a bit like the 1990s computer game, Tetris.

If you’ve never played it (well done you, that’s hours of your life you just got back), the game is to steer awkward-shaped blocks across your screen. As they fall at speed from top to bottom, your job is to direct the Tetris blocks left or right, to make sure they each settle in the right position. And to build a neatly organised layer of blocks forming a wall.

Why is portable digital identity like Tetris??

Because all the capabilities required to build the digital identity ‘wall’ are awkward-shaped blocks. And each need to settle neatly together to make the whole thing work.

The first ‘digital ID’ Tetris blocks started falling around 2015. When some pioneering cryptographers and identity teams began looking at Web3 wallets, and thinking about how?digital identity?could act more like?paper identity.?

More portable. More private. More permissioned.

Those early digital ID Tetris blocks turned out to be?digital identity wallets?and?verifiable credentials.

Over the next few years, more blocks came tumbling down the screen, and started to form a digital identity wall.?

Capabilities like?‘how to create a registry of trusted issuers’. And?‘how to exchange private data using secure protocols’. Later, capabilities were added like?‘how to sync data across different digital wallets’.?And?‘how to structure the data inside verifiable credentials’. Even?‘how to create a?privacy-preserving ‘pass' using verifiable credentials'

Block after block, brick by brick, over the last eight years the portable digital identity wall has grown steadily. And the Tetris blocks continue to fall into place.?

But like the real Tetris game, the blocks are now falling faster and faster. The digital identity and wallet capabilities now available are accelerating. We now have good answers for key management, backup and recovery, and much more.?

But there are a number of missing pieces still.

Commercial models. Discoverability. Distribution. Wallet ‘extensions’ (there’s an exciting post to be written on that one, coming soon).

But one of the most important digital identity ‘Tetris blocks’ to fall into place will be?Wallet?User Experience.

How should verifiable credentials look? What should a wallet experience?feel?like? Or should both wallets and credentials be completely hidden away behind the scenes??

Should credential and wallet interfaces be gesture-based or chat-based? Should they be conversational and ongoing, or transactional? Should they use common UI patterns that are recognised across use cases and sectors (a bit like ‘pinch to zoom’ has become a gesture for digital touch interfaces)? And should there be some kind of credentials ‘trust mark’ across sectors??

The list of UX questions goes on.?

Yes, UX and UI are fertile areas for discovery and design. But they are perhaps one of the most important 'Tetris blocks to get right.?

Because they will be the way we?experience?our digital wallets and Personal AIs.?They will be the way we build digital trust.

So what’s the start of the art for digital wallet user experience? I’m glad you asked.?

This paper is a brilliant summary of the latest thinking and recommendations from the world of digital wallet UX design.

If you’re into the UI/UX of wallets, and reimagining customer experiences, then this one is for you. (For everyone else, you can go back to playing Tetris).

The Power Of DIDs - giving products their own Digital Product Passport

This is a technical one. But super interesting and valuable for anyone working in product management and digital wallets.

It’s the ideas around?Digital Product Passports. These become possible using decentralised identifiers (DIDs), a new and open standard for identifying people, organisations and things. It’s the same technology behind self sovereign identity and digital identity wallets.

Completely different to a product bar code, a ‘product passport’ could connect users to trusted data sources behind the product itself.?

The opportunities are limitless. Imagine linking to user guides, ingredient lists or recipes. To product provenance and the product’s own delivery footprint. Perhaps even to ‘buyers like me’. Or directly to experts who also use the products. It goes on.

The Digital Product Passport is the brainchild of Markus Sabadello, the CEO and founder of DanubeTech. He’s one of the grandparents of the technology behind personal data wallets and portable digital identity (notably the ‘universal resolver’).?

If you want to go deep into the tech behind SSI, and the amazing opportunities that are already being unleashed through Decentralised Identifiers (DIDs), I highly recommend following?Markus’ regular updates.

Apple patents AirPods to measure bio-signals and brain activity?

Apple’s business strategy is well understood: sell beautiful experiences on top of carefully controlled hardware (e.g. Macbooks, Vision Pro headsets), together with locked-down software (e.g. iOS), smart peripherals (e.g. Airpods) and high-margin digital services (iCloud storage).

Each has its own attractive revenue stream. But it’s the combination of products and services that make Apple a money-printing machine.?

In 2020, Apple Airpods alone generated $23bn billion, more than Twitter, Spotify and Square combined.

Apple quietly files patents all the time. Hardware tooling, gesture interfaces, payment processes, even digital identity.?

But there’s a new one worth watching.?Headphones that can measure biometric signals and?brainwaves.

While Elon Musk is busy embedding wet chips into your brain, Tim Cook is going ‘over the top’, adding sensors (and of course AI) into the devices?we already wear.?First watches, now smart in-ear headphones.

I have always been fascinated by intelligent wearables and sensors. For example I’m bullish on?smart jewellery and rings?that can represent verifiable bank cards and other data attributes. Now you really can just ‘touch-to-pay’.?

But the smart clip-ons, sunglasses and headset makers have so far failed to pass the ‘glass-hole’?test.

Brain waves might just be the next massive health data category. Especially when you overlay Personal AI models. Brain signals today are hard to track, and most research has to be done in a lab using fMRIs.

But if a trusted brand came along - especially one positioning itself around privacy and the trusted use of personal data - with an ecosystem of services where a biometric device could fit in neatly, well that could be super interesting.?

I don’t like the idea of more hardware/software customer lock-in. But those brain signals are clearly deeply sensitive health data. And therefore are subject to stringent, existing data protection regulations.?Including data portability under the GDPR.

It will be interesting to see how this plays out, and what role our biometrics will play in Apple’s ecosystem.?

A portable digital wallet, anyone?

AI is clumsy, disrespectful and biased - again

If you ever needed another way to show how GenAI can be brilliant at some things, but deeply clumsy, disrespectful and biased in others, look no further than this recent awful story about an Asian MIT graduate.?

She asked an AI to ‘create a professional headshot’ for her LinkedIn bio.?

Instead,?it changed her eye and skin colour.?And her race.

These tech-meets-culture shock moments matter. Because they are ‘above the waterline’. We can see and feel the real-world impacts.?

The much larger risk here is that we often can’t see the impacts of Personal AI. All the rest of the bias and discrimination embedded in our intelligent tools.?

Because these other shock moments are?invisible. Hidden. Lurking in our chatbot interfaces about money management or employment advice.

You can only trust an AI when you can control it.

That includes knowing much more about the AI training data and the training process. Predictably, we’re once again back to the?difference between ‘Personal’ and ‘Personalised’.

Vitalik Buterin weighs in on biometric proof of personhood

Unless you’ve been living in a cave, you’ll know that?

1. Sam Altman (yes, that one from OpenAI) has finally now launched?WorldCoin
2. It’s getting a frosty reaction from many, especially?data protection authorities around the world who are citing?‘questionable practices’. In Kenya, where apparently Worldcoin adoption has been accelerating, the government has just suspended Worldcoin’s operations

Altman is offering?‘Proof Of Human’?in a particularly provocative way: by scanning your iris.?

And for good measure, WorldCoin then stores a secure version of your biometric data on a blockchain and issues a Web3 token against it.?

The shortest and best explainer I’ve seen is from the?BBC, with the most excellent headline: ‘Sam Altman launches eyeball scanning crypto coin”.

Holy blockchain, tokens and biometrics Batman. 2017 just called and wants its tech-bro news headline back.

Now many are asking: is there a better way? Some folks are pointing to the more scalable and privacy-preserving option of using a?social graph to prove who you are.?The idea is that if 57 people say that I am Jamie, then perhaps that’s a good-enough signal that others can trust.

And it’s way more private than biometrics because?you don’t need to know who those 57 people are. Just that they too have been ‘vouched for’ by others.?

Trusted social graphs come down to needing a set of?root individuals?(‘trust anchors’) who kick things off. An initial group of already trusted people who vouch for each other, and then others in their network. Those others then vouch for others, and so on.?

If done right, a decentralised social graph can avoid the biometric difficulties that Altman is proposing. It’s much harder to get off the ground, but ultimately more private and secure (it’s much harder to hack).

When Vitalik Buterin, the founder of Ethereum, weighs in on a topic, he usually goes pretty deep. Understanding ‘unique proof of personhood’ is no exception, and?he’s just written an excellent piece?about?the perceived tradeoffs around privacy, security and scalability when building ‘personhood’ use cases.

He looks at why it is such a hard problem to solve,?and discusses if social graphs could be a longer-term answer to the short-term trade-offs using biometrics.?

As Sherlock Holmes would say, this is a three-pipe problem. And so Vitalik’s paper is required (and long) reading if you’re into biometrics, digital ID or proof of personhood.

Bonus link: Forrester’s Martha Bennett has done an?excellent breakdown of why WorldCoin won’t work?(“Iris biometrics are clumsy at best and inaccurate at worst”).?

And a final side point: aside from Vitalik’s nonsense around ‘Soul-Based Tokens’,?he’s getting dangerously close?to actually understanding how digital ID could work using a digital wallet and ‘zero knowledge proofs’.

'FraudGPT' is tailored for sophisticated attacks

For years, people have argued that technology isn’t good, nor bad. It’s neutral.?

Perhaps that was true of cars and knives and explosives. But AIs are different: they can learn by themselves, and code and deploy new software technologies on their own.?

I wonder if certain sub-groups of AI platforms will start to be considered good or bad in their own right.?

Exhibit A:?WormGPT, a dedicated platform to launch cyber attacks.

OpenAI, Bard and their family of smart AI platforms have generally put guardrails around them. In order, they say, to protect their users, other businesses and society at large (and themselves of course).?

They define what you can and can’t do. Things you can and can’t ask for. Like advice on suicide or murder for example. But as the recently leaked paper from Google highlighted, these AI models can be mirrored, copied and redeployed easily.?It turns out there is no competitive moat for these open LLMs.

Which means that copycat AI platforms will emerge, and fast. Tailored for certain use cases and domains, like legal or medicine.?But also made available without those guardrails.

Hot on the heels of?WormGPT?(an AI trained specifically to help bad actors) comes?FraudGPT. As?Hackernoon?puts it:

“…the tool could be used to write malicious code, create undetectable malware, find leaks and vulnerabilities, and there have been more than?3,000 confirmed sales and reviews.

“Such tools, besides taking the phishing-as-a-service (PhaaS) model to the next level, could act as a launchpad for novice actors looking to mount convincing phishing and business email compromise (BEC) attacks at scale, leading to the theft of sensitive information and unauthorized wire payments.”

The toothpaste is now definitely out of the tube.?

International trade goes digital - using a smart identifier

The UK just passed a new law - catchily-titled the?Electronic Trade Documents Act?- that’s going to have an outsized impact on the digital identity market.?

It’s a curious corner of law around digitising international trade. And about helping businesses and supply chains become more reliable, faster and more efficient.?

Digitising any traditionally manual, paper-based process is usually a win for everyone involved. But buried in the details of this law is the use of new?digital?identifiers.?

Why should we care about supply chain identifiers? Because the new law will mean businesses can share and accept digital ‘LEIs’, or?digital?Legal Entity Identifiers. And those, my dear friends, will be based on verifiable credential technology. For a deeper dive on that, read my?recent post on LEIs and Organisation ID.

The medium-term knock-on implications could be profound.?

Organisations will be able (and perhaps required) to prove who - and what - they are, seamlessly and instantly across the whole supply chain.?

And if companies can do that with a bunch of other businesses doing trade with each other,?then they should be able to prove who they are to people, too.

This quiet legal update has the potential over the next few years to become an accelerant for the market for legal organisation identity. And once that’s in place, and once customers have their own digital wallets, individuals will be able to check which company they are dealing with online.?

Is it really the bank calling? Is it really the right delivery person? Is it really the online retailer you think it is?

We’ll soon be able to find out.

?? OTHER THINGS

There are far too many interesting and important Customer Future things to include in this edition. So here are some more links to chew on:

• Bloomberg CPMs surged 20% after ditching third-party programmatic ads?READ
• Zombie pixels, data consent and de-identified user tracking face the heat in the Australian data broking probe?READ
• Decentralized identity in finance?WATCH
• What would the internet of people look like now??READ
• Empowering Private Personalisation: Customer-Centricity in the Age of AI?READ
• Wallet Wars or Collaborative Wallet Ecosystems??READ
• OpenAI's trust and safety lead is leaving the company?READ
• Why loyalty and trust are not the same thing?READ
• Will "Customering" replace Marketing??LISTEN

And that’s a wrap. Stay tuned for more Customer Futures soon, both here and over at?www.customerfutures.com.