Category: Me2B

Markets as conversations with robots

From the Google AI blogTowards a Conversational Agent that Can Chat About…Anything:

In “Towards a Human-like Open-Domain Chatbot”, we present Meena, a 2.6 billion parameter end-to-end trained neural conversational model. We show that Meena can conduct conversations that are more sensible and specific than existing state-of-the-art chatbots. Such improvements are reflected through a new human evaluation metric that we propose for open-domain chatbots, called Sensibleness and Specificity Average (SSA), which captures basic, but important attributes for human conversation. Remarkably, we demonstrate that perplexity, an automatic metric that is readily available to any neural conversational models, highly correlates with SSA.

A chat between Meena (left) and a person (right).

Meena
Meena is an end-to-end, neural conversational model that learns to respond sensibly to a given conversational context. The training objective is to minimize perplexity, the uncertainty of predicting the next token (in this case, the next word in a conversation). At its heart lies the Evolved Transformer seq2seq architecture, a Transformer architecture discovered by evolutionary neural architecture search to improve perplexity.
 
Concretely, Meena has a single Evolved Transformer encoder block and 13 Evolved Transformer decoder blocks as illustrated below. The encoder is responsible for processing the conversation context to help Meena understand what has already been said in the conversation. The decoder then uses that information to formulate an actual response. Through tuning the hyper-parameters, we discovered that a more powerful decoder was the key to higher conversational quality.
So how about turning this around?

What if Google sold or gave a Meena model to people—a model Google wouldn’t be able to spy on—so people could use it to chat sensibly with robots or people at companies?

Possible?

If, in the future (which is now—it’s freaking 2020 already), people will have robots of their own, why not one for dealing with companies, which themselves are turning their sales and customer service systems over to robots anyway?

People are the real edge

You Need to Move from Cloud Computing to Edge Computing Now!, writes Sabina Pokhrel in Towards Data Science. The reason, says her subhead, is that “Edge Computing market size is expected to reach USD 29 billion by 2025.” (Source: Grand View Research.) The second person “You” in the headline is business. Not the people at the edge. At least not yet.

We need to fix that.

By we, I mean each of us—as independent individuals and as collected groups—and with full agency in both roles. The Edge Computing is both.

The article  illustrates the move to Edge Computing this way:

The four items at the bottom (taxi, surveillance camera, traffic light, and smartphone) are at the edges of corporate systems. That’s what the Edge Computing talk is about. But one of those—the phone—is also yours. In fact it is primarily yours. And you are the true edge, because you are an independent actor.

More than any device in the world, that phone is the people’s edge, because connected device is more personal. Our phones are, almost literally, extensions of ourselves—to a degree that being without one in the connected world is a real disability.

Given phones importance to us, we need to be in charge of whatever edge computing happens there. Simple as that. We cannot be puppets at the ends of corporate strings.

I am sure that this is not a consideration for most of those working on cloud computing, edge computing, or moving computation from one to the other.

So we need to make clear that our agency over the computation in our personal devices is a primary design consideration. We need to do that with tech, with policy, and with advocacy.

This is not a matter of asking companies and governments to please give us some agency. We need to create that agency for ourselves, much as we’ve learned to walk, talk and act on our own. We don’t have “Walking as a Service” or “Talking as a Service.” Because those are only things an individual human being can do. Likewise there should be things only an individual human with a phone can do. On their own. At scale. Across all companies and governments.

Pretty much everything written here and tagged VRM describes that work and ways to approach that challenge.

Recently some of us (me included) have been working to establish Me2B as a better name for VRM than VRM.  It occurs to me, in reading this piece, that the e in Me2B could stand for edge. Just a thought.

If we succeed, there is no way edge computing gets talked about, or worked on, without respecting the Me’s of the world, and their essential roles in operating, controlling, managing and otherwise making the most of those edges—for the good of the businesses they deal with as well as themselves.

 

 

Personal scale

Way back in 1995, when our family was still new to the Web, my wife asked a question that is one of the big reasons I started ProjectVRM: Why can’t I take my own shopping cart from one site to another?

The bad but true answer is that every site wants you to use their shopping cart. The good but not-yet-true answer is that nobody has invented it yet. By that I mean: not  a truly personal one, based on open standards that make it possible for lots of developers to compete at making the best personal shopping cart for you.

Think about what you might be able to do with a PSC (Personal Shopping Cart) online that you can’t do with a physical one offline:

  • Take it from store to store, just as you do with your browser. This should go without saying, but it’s still worth repeating, because it would be way cool.
  • Have a list of everything parked already in your carts within each store.
  • Know what prices have changed, or are about to change, for the products in your carts in each store.
  • Notify every retailer you trust that you intend to buy X, Y or Z, with restrictions (meaning your terms and conditions) on the use of that information, and in a way that will let you know if those restrictions are violated. This is called intentcasting, and there are a pile of companies already in that business.
  • Have a way to change your name and other contact information, for all the stores you deal with, in one move.
  • Control your subscriptions to each store’s emailings and promotional materials.
  • Have your  own way to express genuine loyalty , rather than suffering with as many coercive and goofy “loyalty programs” as there are companies
  • Have a standard way to share your experiences with the companies that make and sell the products you’ve bought, and to suggest improvements—and for those companies to share back updates and improvements you should know about.
  • Have wallets of your own, rather than only those provided by platforms.
  • Connect to your collection of receipts, instruction manuals and other relevant information for all the stuff you’ve already bought or currently rent. (Note that this collection is for the Internet of your things—one you control for yourself, and is not a set of suction cups on corporate tentacles.)
  • Your own standard way to call for service or support, for stuff you’ve bought or rented, rather than suffering with as many different ways to do that as there are companies you’ve engaged

All of these things are Me2B, and will give each of us scale, much as the standards that make the Internet, browsers and email all give us scale. And that scale will be just as good for the companies we deal with as are the Internet, browsers and email.

If you think “none of the stores out there will want any of this, because they won’t control it,” think about what personal operating systems and browsers on every device have already done for stores by making the customer interface standard. What we’re talking about here is enlarging that interface.

I’d love to see if there is any economics research and/or scholarship on personal scale and its leverage (such as personal operating systems, devices and browsers give us) in the digital world). Because it’s a case that needs to be made.

Of course, there’s money to me made as well, because there will be so many more, better and standard ways for companies to deal with customers than current tools (including email, apps and browsers) can by themselves.

A positive look at Me2B

Somehow Martin Geddes and I were both at PIE2017 in London a few days ago and missed each other. That bums me because nobody in tech is more thoughtful and deep than Martin, and it would have been great to see him there. Still, we have his excellent report on the conference, which I highly recommend.

The theme of the conference was #Me2B, a perfect synonym (or synotag) for both #VRM and #CustomerTech, and hugely gratifying for us at ProjectVRM. As Martin says in his report,

This conference is an important one, as it has not sold its soul to the identity harvesters, nor rejected commercialism for utopian social visions by excluding them. It brings together the different parts and players, accepts the imperfection of our present reality, and celebrates the genuine progress being made.

Another pull-quote:

…if Facebook (and other identity harvesting companies) performed the same surveillance and stalking actions in the physical world as they do online, there would be riots. How dare you do that to my children, family and friends!

On the other hand, there are many people working to empower the “buy side”, helping people to make better decisions. Rather than identity harvesting, they perform “identity projection”, augmenting the power of the individual over the system of choice around them.

The main demand side commercial opportunity at the moment are applications like price comparison shopping. In the not too distant future is may transform how we eat, and drive a “food as medicine” model, paid for by life insurers to reduce claims.

The core issue is “who is my data empowering, and to what ends?”. If it is personal data, then there needs to be only one ultimate answer: it must empower you, and to your own benefit (where that is a legitimate intent, i.e. not fraud). Anything else is a tyranny to be avoided.

The good news is that these apparently unreconcilable views and systems can find a middle ground. There are technologies being built that allow for every party to win: the user, the merchant, and the identity broker. That these appear to be gaining ground, and removing the friction from the “identity supply chain”, is room for optimism.

Encouraging technologies that enable the individual to win is what ProjectVRM is all about. Same goes for Customer Commons, our nonprofit spin-off. Nice to know others (especially ones as smart and observant as Martin) see them gaining ground.

Martin also writes,

It is not merely for suppliers in the digital identity and personal information supply chain. Any enterprise can aspire to deliver a smart customer journey using smart contracts powered by personal information. All enterprises can deliver a better experience by helping customers to make better choices.

True.

The only problem with companies delivering better experiences by themselves is that every one of them is doing it differently, often using the same back-end SaaS systems (e.g. from Salesforce, Oracle, IBM, et. al.).

We need ways customers can have their own standard ways to change personal data settings (e.g. name, address, credit card info), call for support and supply useful intelligence to any of the companies they deal with, and to do any of those in one move.

See, just as companies need scale across all the customers they deal with, customers need scale across all the companies they deal with. I visit the possibilities for that here, here, here, and here.

On the topic of privacy, here’s a bonus link.

And, since Martin takes a very useful identity angle in his report, I invite him to come to the next Internet Identity Workshop, which Phil Windley, Kaliya @IdentityWoman and I put on twice a year at the Computer History Museum. The next, our 26th, is 3-5 April 2018.

 

 

© 2020 ProjectVRM

Theme by Anders NorenUp ↑