Introduction

The Biggest Opportunities Created by Chatbots and IVAs

I.1 © AI BaaS UG, Munich, 2019. Diagram of our proprietary algorithm VAIP [Virtual Artificially Intelligent Patois]

The UK Government’s Multibillion Pound Bot Budget and Fears of the “Digital Poorhouse”

A Guardian investigation has established that 140 councils out of 408 have now invested in the software contracts, which can run into millions of pounds, more than double the previous estimates. The systems are being deployed to provide automated guidance on benefit claims, prevent child abuse, and allocate school places. But concerns have been raised about privacy and data security, the ability of council officials to understand how some of the systems work, and the difficulty for citizens in challenging automated decisions.1

What has this got to do with chatbots or Artificial Intelligence? As it turns out, the embodiment of these “automated systems and software contracts” is going to be 2D online Virtual Assistants who will converse 24/7 with the citizens they are meant to serve. Chatbot Public Servants, no less! And the systems referred to actually use Machine Learning or AI tech to automate their number crunching and processing of people’s data.

Significantly, British government’s reported budgets for just this one application or use case has been—up until the time of writing this book, which is literally “Brexit eve”—in the multiple billions of pounds, not just millions of sterling. Here is a quick tally of the numbers exposed in the investigative journalism of the left-wing newspaper and media ­platform, The Guardian in London:

• The Department for Work and Pensions (DWP) hired around 1,000 IT staff over 18 months.
• In that time, they increased spending to about £8 million per annum on an “intelligent automation garage” to develop over 100 welfare robots; 16 are already communicating with claimants of welfare payments, according to The Guardian.
• The DWP spokesperson interviewed described the overall departmental, government budget as “£95 billion for a compassionate safety net creating a digital service that suits the way most people use technology.”
• The official, openly accessible, published DWP Digital Budget has risen by 17 percent to £1.1 billion over a 12-month period.2

The Guardian’s special investigation—which I discuss again in the “Intrapreneurship” section of Chapter 5 and in the Conclusion of this book—was echoed by the right-wing tabloids after the leftist reporters broke the news. The Mirror followed The Guardian’s headline of “March of the ‘Welfare Robot’ Triggers Fear for Poorest” and its editorial comparing the British “digital poorhouse” to the negative, democracy threatening developments in the United States, India, and Australia.3

As The Guardian explained on its front page, this British “Ministry” or federal department for social welfare payments and retiree’s public pensions had engaged foreigners (mostly Americans) to run these systems that would create a “digital identity” or check your profile online to detect fraud and (over)payments. “As well as contracts with the outsourcing multinationals IBM, Tata Consultancy and Capgemini, it is also working with UiPath, a New York-based firm co-founded by Daniel Dines, the world’s first ‘bot billionaire’ who last month said: ‘I want a robot for every person.’”4

I discuss this UK example again in the Intrapreneurship section about the public sector, the Government Guidelines and Checklists. Suffice to say at this point, the confusion around the terms “robot,” “chatbot,” AI, and automation are evident in this tabloid story. When you scroll to the bottom of The Mirror report, you discover that actually no “robots” nor chatbots for that matter have actually been deployed by the DWP. Not a single member of the public has used the Conversational AI interface of the American billionaire: “A 2018 blog by DWP senior product owner Shaun Williamson said officials were ‘exploring the potential of chatbots’—claiming they could cut calls about sickness benefit by 200,000 per week. A DWP spokeswoman said there are not currently any claimant-facing chatbots in the system.”5

This is a clear example of the hype and fears surrounding chatbot deployment. Anxieties seep from the public domain into the private homes of the collective “user,” spurred on by the media bandying about misconceptions even misinformation within a single article. It is important to stress though their overarching ethical, human rights concerns about data protection and privacy are indeed be valid. Too often however, the commercial outcome is that the private sector providers of Bots as a Service (BaaS), are left high and dry in the confusion.

I.2 © Cliff Lee, Devon, 2019. Prototype of a chatbot newsreader presenting the news from the European Union’s RSS news feeds in Australia. This botification was too early to market as users/readers did not even know what a chatbot was in 2011–2012, let alone how to use it. Nor were chat apps in existence, it could only be trialled online via a website

I do not wish to discount the genuine fears and examples of the automation system “gone wrong,” where according to the aforementioned journalists, claimants have died from hunger (in India), suicided, or had mental breakdowns ending in homelessness (in the UK and the United States). That is truly appalling and inexcusable when “the system” lets them down in such tragic, inhumane ways. The United Nations, at the time of writing, was just about to release an alarming report on “UN studies of poverty in the UK and US, as well as submissions from experts and governments from 34 countries,” warning of the dangers of “digitising social protection…addressing the harassment, targeting and punishment of those living in the digital poorhouse.”6

Yet I believe that it is infinitely worse for welfare recipients and vulnerable elderly or disabled to be caught in human-caused backlogs that delay their payments because of staff inefficiencies and loss of oversight by managers. With the media ready to “get on their case,” we must be confident that government bodies like the DWP will have “real people” available to “solve complex cases” once the (chat)bots have flagged urgent and/or life-threatening situations, as the officials are indeed promising in the following. For the sake of balanced reporting, the Guardian’s social affairs correspondent included “the Government’s response” by a DWP spokesperson who basically makes my argument mentioned earlier:

The DWP said humans remained available to help: “We continue to invest in frontline colleagues, from phone lines to work coaches to front-of-house staff,” said a spokesperson. “This means people who struggle with digital services, or are worried about a wrong decision, can get the help they need.”7

The massive British public sector budgets for this one “vertical”—social welfare and pensions—and its use cases within just one governmental department demonstrate conclusively that chatbots are here to stay. In 2016, the well-known, American financial news platform Business Insider (though it’s now German-owned by Axel Springer), predicted that by this year, 80 percent of businesses globally would have, be planning for, or want a chatbot integrated into their back end CRMs and/or public facing systems.

The survey included responses from 800 decision makers including chief marketing officers, chief strategy officers, senior marketers, and senior sales executives from France, the Netherlands, South Africa, and the UK. When asked which emerging technologies they are already using and which they intended to implement, 80 percent of respondents said they already used or planned to use chatbots by 2020.8

This survey was actually conducted by Oracle, who incidentally ended up releasing their own “state-of-the-art” chatbot across platforms for their high-end CRM system. Oracle’s survey— that made headlines worldwide four years ago in the mainstream media with their 80 percent of botified or about to botify companies claim—predicted that the automation triggered by chatbot deployments would inevitably transform the economy:

Twenty-nine percent of customer service positions in the US could be automated through chatbots and other tech, according to Public Tableau. We estimate this translates to $23 billion in savings from annual salaries, which does not even factor in additional workforce costs like health insurance.9

Their predictions were based on this remarkable graph; its data sources were McKinsey consultancy and the U.S. Office of Personnel Management.

I.3 © Tania Peitzker on gravatar.com 2012-2019. Here you can see many of the 2D chatbot pilots my former venture velmai undertook as unpaid Greenfield projects. Expectations of quick uptake of bots were strong yet users and therefore prospective clients hesitated to adopt this new technology for sales, marketing and communication. (accessed https on November 3, 2019)

Well, it is the year 2020. Did this shift actually happen? If not, why not? Needless to say, there has been a lot of discussion in that time about the social impact of potential and actual job losses. This is part of the general hype around the words AI and Artificial Intelligence. Known often as scaremongering in the media, along the lines that “the robots are taking our jobs” or else prevalent in science fiction narratives with the ever familiar, existentially threatening plot that “AI will replace you.”

Chatbots Set to Be a Trillion Dollar Market

MarketsandMarkets has published lengthy reports on my particular industry niche; mostly about chatbots, Intelligent Virtual Assistants (IVAs), and voice-based cognitive interfaces being worth over $1 trillion by 2025. From 2017 onward, I have been interviewed four times in depth, recorded for over an hour by a team of business intelligence (BI) analysts in India, plus filled out their detailed survey postinterview. This was as a source for their reports, priced at $8,000+ each on IVAs, the Cognitive Market, Conversational AI, and Conversational Commerce.10

As an introductory reply to my own rhetorical question, did the Business Insider prediction come true, consider these points:

• Yes, there has been a significant multiplication of social media chatbots and their corporate adoption. Just search online for “chatbots” or Virtual Assistants and you will see endless pages and ads about thousands of bot developer companies and avatars.
• There are now countless number of corporate bots and use cases such as the Unilever chatbot and Archant media news archive bot by Google Ubisend in the UK; only a few years ago, there were just a few hundred chatbots you could point to online.
• They have exploded in numbers by being hosted on frameworks or APIs that give templates to bot developers at all levels, 2D ones only like wit.ai and open.ai enhancing independent bot development in most languages globally.
• In 2020, we now see novel use cases like Disney’s Zootopia bot cooperating with children and the interactive detective character to solve a mystery and Marvel’s “Guarding the Galaxy” character.11
• Other digital media firms have created “choose your own ending of films,” children’s novels, or else interactive computer gaming with a bespoke chatbot character.12
• ChatFuel and thousands of 2D chatbot companies are releasing specific use cases like “Do it Yourself Drag and Drop Ada,” the UN Refugee survey to support asylum seekers lodge their claim via their smartphone once they arrive at their destination country of refuge, various law bots to support solicitors rather than the client, recipe bots on Skype, and chatbot services like real estate, bookkeeping, messaging for team work in Slack, Kik, Telegram, and WeChat.
• Yes 2020 does seem like a trillion dollar market! There are definitely more chatbots than ever in the B2C backend systems of major SaaS providers like SAGE (Peg), Accenture (Amy), and Publicis (Marcel) not to mention Oracle—see their persuasive video about their AI bot capabilities at the end of this section in the chapter.
• Yes, there are more CRM and enterprise solution providers of chatbot systems, for example, www.recast.ai in France, created by a couple of young women in a French incubator. This company was sold to the famous German CRM SaaS multinational corporation SAP for an estimated €40 million in 2018!
• Such M&As—for example Bloomsbury AI from London was a recent acquisition by a U.S. firm reported to be valued at about $30 million,13—have encouraged the latest generation at secondary school, school leaver, and tertiary level coders and programmers to create more applied chatbot tasks with Pizza Hut and CNN bots seeming like the tip of the iceberg, as were the early days of the pioneering Dutch KLM and Alaska Airlines with their inaugural industry pilots of 2D chatbots to assist passengers online with flight bookings and information, run on the respective airlines’ branded booking websites.
• Yes, the forecast looks correct, as IBM Watson endeavors to white label much of its data mining software and deep learning features of its algorithm as botified cognitive interfaces, for example, I came across a white-labeled Watson Health application from India recently, selling into the Australian market; ditto for Microsoft with its Bot Network featuring the proposed Master Bot Cortana, though she still keeps a pretty low profile, that is, no evident mass consumer uptake of the Microsoft desktop chatbot.
• Deutsche Bahn surprised my tech venture velmai/AI BaaS with the release of SEMMI in Berlin Central Station a month after we trialed our Amalia 3D AI bot hologram in a mall in Cologne. SEMMI’s hardware was in fact a Swedish robot head displaying a hologram face.
• Her facial hologram could blink and show some expressions, yet the actual robotic head could not move much, only rotate from side to side. It appears to have had excellent speakers to hear travelers’ questions about timetables and trips with the German rail system.
• Like our Amalia I, SEMMI could also direct people to the toilets and cafes in response to the question “where can I get a coffee,” the weather, or exits. Though we think our prototype was more advanced in terms of content, Amalia I not only gave directions, she also provided upon request detailed descriptions for over 30 retailers, restaurants, and services in our shopping center.
• Amalia I was more conversational than SEMMI appears to be. For instance, Amalia proactively engaged with shoppers by asking them how their day was, perhaps they needed to buy a gift for someone in their family (then making suggestions for various retailers in the mall) as well as soft selling the idea the person should get a present for their mother, just to say you appreciate her anytime of the year!
• Like SEMMI by Deutsche Bahn, Amalia I was only trialed publicly for just on four weeks before we took her “back to the lab” to improve her hardware and tweak her software.
• Fortunately we escaped a public flop like DB had to endure, see the video reportage by the German tabloid Bildzeitung. The brief mishap—SEMMI did start working perfectly after a “false start” —went viral somewhat on chat apps like Twitter retweeting blog posts about it, as well at RT (Russia Today) doing an extensive video reportage on the supposed failure, even though the footage showed journalists successfully communicating with the robot head after 10 minutes or so, in both German and English.
• The negative “press” pilloried the poor computer scientists employed by Deutsche Bahn because their creation became silent and refused to answer for a while! Not long, only five or ten minutes, but that was long enough to cause publicity damage to the new product launch in this emerging tech, which seems like an easy target to get a lot of hits if you write an exciting, amusing story about the innovation with a great deal of Schadenfreude and few facts.

Unforeseen (positive) Consequences of Chatbot Growth

Where the Business Insider forecast did not materialize or else did not predict the evolution:

• Oracle market researchers, McKinsey consultants, BI analysts, and the U.S. Department officials did not see that chatbots would be morphing into (a) IVAs, now known as AI-powered bots or AI bots
• That IVAs would be further developed into voice-based bots
• And that companies with agile proprietary algorithms like mine would end up pivoting from 2D chatbot production to 3D cognitive avatars, thereby challenging the two main U.S. tech giants in this space, Google and Amazon, with Mixed Reality AI bots like holograms and AR integrating chatbots or IVAs
• That challengers to the tech giants, upstarts as they are called, would also be more agile and able to customize the various personalized user interfaces
• A new term would come into its own: Cognitive Interfaces would be the next-gen chatbots leaving the 2D static rule-based earlier versions somewhat behind.
• Consequently, legacy pioneers like Nuance Communications, Pandora Chatbots, and Chatfuel, not to mention a plethora of thousands of bot developers who can “only” create 2D online chatbots
• Proprietary algorithms rather than open source and bot platforms with APIs would lead this cutting edge of the experience economy run by voice or speech recognition; indeed a new term has entered the lexicon, CxO or Customer Experience Officer! The overarching executive who makes sure the customer is experiencing the product or service in the best, most enjoyable, and memorable way.

The Biggest Risks of Voice Bots (Chatbots With Speech Recognition)

1. Can be hacked leading to people as well as their vehicles being manipulated via IoT technology. See the frequently cited lack of cybersecurity in smart appliances and toys, which the German government was one of the first to legislate for enforcing better industrial design and above all cybersecurity from the manufacturers.
2. Can be transcribed and listened in to as Amazon and Google have been “caught” listening to their customers, with Der Spiegel and other media revealing that both tech giants have illegally stored and analyzed data or conversations that were not meant to be heard let alone archived and analyzed.
3. Particularly insidious is that even music can contain hacker messages to the IoT devices to give instructions unheard by the user, for example, transfer money, open doors, and garages and give info.
4. Voice impersonators with people’s voices imitated or in fact stolen; this has already led to this new type of stolen “digital password” stored in rogue countries in voice banks, after fraudsters have hacked into an IoT home or infrastructure deployment.

Nevertheless, Ivy League researchers who looked into psychology and Internet users’ well-being have spoken out that the devices are creating addiction due to people retreating into solitude with them and their online interaction only. In a Yale University dialogue recorded for the Coursera MOOC course on “The Science of Well-Being” by Professor Dr Laurie Santos of Yale’s Psychology Faculty, she interviews Professor Nicholas Epley of the University of Chicago Booth School of Business.14 Epley unexpectedly suggests that adding voice to all our devices would alleviate this antisocial phenomenon. Applied to his “real life experiments on Chicago commuter trains,” it could force smartphone and tablet users to begin interacting again socially by using their own voices and hearing the voice of others.15 An extension of Epley’s insight, I think is that the voice added to the device could be botified or human in order to bring about a global, revolutionary and fundamentally healthy use of the Internet manifested in its social media form and smartphone connectivity.

Overarching Risks of AI Bot Technologies

• There is a risk and danger of being overwhelmed by too many choices. See the plethora and abundance of search results when you use just a few keywords like “chatbots” or “Intelligent Virtual Assistant.”
• However, so many independent bot developers without provenance could mean trouble—who is behind the code, is it robust against hackers, has it been partly programmed by hackers with “doors” they can open at a later date? Is malware hidden inside your bot?
• As the EU Department for Competition has already started fining, there is a risk of seeming cartels leaving the field of enterprise solutions to just a few players like Google, Amazon, Samsung, Apple, Microsoft, and IBM. If there is a lack of voice-based BaaS at a sophisticated level, then only this small group will control the pricing and also control the services available at what conditions. This clearly is not an issue for the 2D chatbot scene or the less advanced IVAs.
• Look at how the Facebook monopoly over Messenger bots caused chatbot entrepreneurs to go out of business during the Cambridge Analytica expose. We lost a big client in Munich as we could not access the demo bot we had prepared for the contract signing. I discuss this in Case Study #10 in this book. Monopolies like Amazon and eBay mean they can cut off satellite businesses without much reproof or redress.
• Privacy and data protection, which I will go into detail in the section on the GDPR.
• Poor coding that allows infiltration and manipulation with offensive libelous speech and publishing.
• Lack of enforcement for breaches of these software standards, no uniform legislation.
• See my VentureBeat articles16 from 2016, which foreshadow core issues that are now magnified in the year 2020:

1. Hate speech has given rise to the Far Right, racism, red baiting as a type of uncensored “Reddit war” on liberal intellectuals; as Chapter 2 examines, reactionary policies are reinforced with armies of chatbots spamming social media accounts (pro-Trump bots, for instance). My post on Microsoft Tay’s Hate Speech is still being read today on VB; I discuss this case study in depth in the next chapter.17
2. Fringe views of the Far Left and the Far Right, not to mention of terrorists, can be disseminated more rapidly with Twitter bots. Political entities like Russians and North Korean propaganda campaigns are utilizing them; China has been accused of the same re. State-run censorship, surveillance, and intimidation.
3. See my VentureBeat post on the “Swiss chatbot arrest,” so on the issues of control and ownership as well as liability for what a bot says and does.18
4. As I discuss later in this book, the “provenance” of advanced bots versus simple bots is crucial to performance as well as cybersecurity. When they truly interact and learn as AI should through conversations with human, then that is the reason why bots are disruptive.
5. As my key VB article also pointed out, it is the main reason we have been blocked by the advertising industry: top-performing bots produce the most transparent, cost-effective return on investment with absolute clarity of metrics.19

Overarching Benefits of Commercial “Botification” Use Cases

• The reason there are hundreds of thousands of bot developers is because businesses want them!
• They work! They are adding value to the enterprise—its branding awareness, sales and marketing campaigns—and improving the bottom line with direct sales and cost savings through more productive staff.
• Despite the past decades of simplistic, “legacy,” manually coded chatbots, who have now been overtaken by the smarter, more versatile and cognitive AI bots, the evolution has carved out a path for the incumbent next-gen AI bots with voice. It has propelled improvements to the 2D hybrids of machine and humans interacting to improve customer service and satisfaction in the Experience Economy for the post-Millennial generation.
• Big brands and big corporations are creating their own cross-platform bespoke bots inhouse because they add value B2B and internally for staff relations, employee contentment, and stakeholder interest and involvement.
• Banks are now adopting them after initially rejecting them; they trust their performance now after many greenfield projects and pilots by them and competitors. Often these trials were done internally for staff to use so the banks’ customers never interacted with these first generation chatbots.
• Fintech in the form of financial services and more transparent credit checks—under GDPR you have the potential to get your personal data more quickly and reliably asking a chatbot instead on an online contact form or phone call, ditto if you want your private info deleted.
• Real estate and intensive service bots are converting more leads to sales.
• FAQs for customer service use cases, sales and marketing for all verticals and industries at scale are working! See the British Ubisend’s myriad of published case studies on their corporate website as just one example, or LivePerson, CreativeVirtual, PandoraBots, ChatFuel, Inbenta, Watson, Artificial Solutions and so on.
• Mental health bots to tackle depression epidemic and morbid obesity, dieting, and welfare of youth to stop cyberbullying, self-harm, suicide.
• See Case Study #5 about an (anonymized) embassy or diplomatic mission wanting to curb hate speech on its Facebook page run from London.

I.4 Here the robot “Pepper” is deployed in a Home Environment and so becomes an “anti-loneliness measure” purchased by a son for the mental health of his solitary mother in Tokyo. This “real scene with real people not actors” was recorded in a German documentary which follows the human-machine relationships in this family, in parallel to a Road Trip love story involving another robot and human American (see Chapter 5 for illustrations). Taken from the film “Hi, AI” by Isa Willinger © Kloos & Co Medien, Berlin, 2019

---