OU blog

Personal Blogs

Chris Colquitt

Artificial minds - An interview with Tony Birch (Shipley Ltd) on theory of mind, executive summaries and 'Rose'

Visible to anyone in the world

This week, I sat down with Tony Birch, MD of Shipley Limited, to discuss their Executive Summary support tool 'Rose' and how it can help salespeople to craft better, more empathetic executive summaries in sales proposals.

 Tell us a little more about Rose

 Rose works predominantly with salespeople to help them in creating an early draft of the Executive Summary(1). It does this by prompting them through a series of questions, rather than asking them to sit down and write on a blank piece of paper. When there's no structure, it can be really hard to write a powerful document like the executive summary. Using these prompts, Rose partially automates information into Shipley's 'four-box' template to help create a persuasive and exciting executive summary.

To what extent does ROSE try to apply theory of mind(2) practices in writing an executive summary? What elements does it try to avoid?

 Rose doesn't directly try to present a theory of mind. There's a good chance that systems like Rose 'trying to be a person' will come eventually, but it's not there yet.

 There are subtle developments that can help systems like Rose in being more accessible and more lifelike, such as voice interaction and mimicking human behaviours. The next generation of something like Rose may be driven by voice and, whilst it's already trying to 'have a conversation' in the passive sense, there's good opportunity to develop this into a more literal conversational process.

Ultimately, Rose provides a format to prepare an executive summary that will allow the decision makers to select you and your solution. It does this through asking questions that guide the writer and helping them to consider the buyer, anticipating what questions they may ask. If you did this early enough, Rose could potentially listen and collate, providing near real-time updates along the way. There are often complaints [in the buying cycle] that people didn't have a conversation. More often, it's a case that the seller or the buyer didn't know how to articulate that information back to the writer. Rose helps to bring these elements together.

John Searle wrote a famous essay about his Chinese Room(3) where he could be passed instructions in Chinese from the outside and, following a set of written instructions, could respond and make it look like he could speak Chinese – though he’d always just be following instructions. His argument is that AI will always just be following instructions and never have a good ‘theory of mind’. Do you think attempts like Rose support his argument and are limited to their programming?

There's been a number of attempts to introduce more empathy into these types of processes, and many of them are just algorithms and playing back words. However, it turns out we can be pretty impressed by these! Sometimes we don't need to be the best - we just need to be better than the rest. Based on our experience at Shipley, we tend to see there are a finite number of reasons why people want to buy ('hot buttons'). Usually less than 40.

Rose sits in the middle of these viewpoints [weak AI and strong AI4], supporting empathy by providing a library of hot buttons for salespeople to choose from. Ultimately, it can help to prompt empathy and alignment - but this is something that good salespeople tend to do. Ultimately it helps to reduce accusations of 'you didn't listen to me' inside the process.

When was the last time AI surprised you by tricking you into thinking it had a theory of mind (Alexa, Google, chatbots, etc.)

I recently ended up searching on a set of symptoms [ed. Nothing serious, hopefully!] and ended up on an American 'WebMD' style site talking to, what I assume, was a chatbot(5). It was obviously working down a decision tree, but prefixed a few pieces of advice with 'now, you're not going to like this'. I thought that was really clever. We'd like to integrate those types of natural language algorithms(6) into proposal automation one day.

Finally, to get the most empathetic executive summary, should salespeople start it with 'we'd like to thank [customer] for inviting us to tender'?

No. Never - unless it's culturally dictated by the region you're tendering in. Ultimately, everyone involved in your process will read the first line of the executive summary. Do you want everyone to know that you're pleased to be invited? Or do you want everyone to know how you're going to help them?

 Calls to action and next steps

  1. Does empathy matter in an executive summary? A study or literature review on the effectiveness of salespeople as correlated to empathy would be welcomed (for example, see Dawson et al (1992), Pilling et al (1994), McBane (1995)) from an interested APMP member or Business School student.
  2. How is theory of mind being integrated (or avoided) in AI, especially as it pertains to corporate sales? We welcome views from technology companies and sales enablement firms on this process. We'd be particularly interested in any attempts to pair guided writing with chatbot and machine learning technologies.

About the contributors:

Tony Birch is founder and Managing Director of Shipley Limited, and a Fellow of the APMP and a Certified Professional under the APMP Certification Programme. As one of the founders of APMP UK, Tony’s mission is to ensure that the role of a bid or proposal manager is recognised throughout the world as a profession - and not just a job. You can find out more at https://www.shipleywins.co.uk/bid-tools

Chris Colquitt is Manager of Global Proposal Management at Clarivate Analytics and a student in Psychology, currently studying with the Open University in the UK.

You can hear the recording of the interview at this link

Notes:

1 - Executive Summary

For readers from outside of the sales, bidding and proposal world - the Executive Summary is a short chapter in a sales document, usually towards the front, that explains the key aspects of your solution or offer, and why the buyer should select you. Good executive summaries will explain the buyer need alongside the solution you are presenting. Poor executive summaries will tend to be one-sided, concentrating on the features of a company's own products rather than applying their benefits to the buyer. For example, think of buying a new laptop and the salesperson only talking in terms of GB Ram and GHz clock speeds, rather than asking you if you want to use it for watching films and checking emails vs. editing photos and playing games.

2 - Theory of Mind

Theory of mind is defined as: 'the ability to attribute mental states—beliefsintentsdesiresemotionsknowledge, etc.—to oneself, and to others, and to understand that others have beliefs, desires, intentions, and perspectives that are different from one's own.' From <https://en.wikipedia.org/wiki/Theory_of_mind>

 Note: For this article, empathy and theory of mind are used largely interchangeably. However there are subtle differences.

3 - Chinese Room

Searle, J. (1999) 'Chinese room argument' in Wilson, R. A. and Keil, F. C (eds) The MIT Encyclopedia of the Cognitive Sciences, Cambridge, MA and London, MIT Press. See a summary at Chinese room

4 - Weak AI and strong AI

Weak AI (or narrow AI) is AI that is focused on a single narrow task, irrespective of how complex that individual task is. Most current AI operates as weak AI. A summary is available at Weak AI

Strong AI (or Artificial General Intelligence) is AI that could successfully replace a human being in a specific task. As such, strong AI needs to include cognitive capabilities such as empathy, problem framing, imaginative problem solving and more. A summary is available at Artificial general intelligence

5 - 'What, I assume, was a chatbot'

The fact that this statement is an assumption means we're already incredibly close to passing the Turing Test, one of the previously assumed barriers between weak and strong AI. See a summary of the Turing Test at Turing test

6 - Natural language

You may be interested in seeing Google's latest use of natural language in AI which stunned me. See a summary at https://www.theverge.com/2018/5/8/17332070/google-assistant-makes-phone-call-demo-duplex-io-2018


Permalink Add your comment
Share post

This blog might contain posts that are only visible to logged-in users, or where only logged-in users can comment. If you have an account on the system, please log in for full access.

Total visits to this blog: 3846