Uploaded byAustin Beer
Designing Intelligent Coversations: A Chatbot Workshop
The document outlines a workshop at SXSW 2017 focused on designing intelligent conversations through chatbots. It covers the basics of chatbots, effective conversation design, and the role of AI and machine learning in creating engaging user interactions. It emphasizes the importance of understanding user relationships and designing conversations that align with business goals.
More Related Content
Similar to Designing Intelligent Coversations: A Chatbot Workshop
![[TechWorldSummit Stockholm 2019] Building Bots for Human with Conversational ...](https://cdn.slidesharecdn.com/ss_thumbnails/building2019stockholmbotsforhumanwithconversationalinterface-190604131903-thumbnail.jpg?width=640&height=640&fit=bounds)
Designing Intelligent Coversations: A Chatbot Workshop
- 1. Hello.45 Main St3rd Fl . Brooklyn, NY 11201 P: 718-625-4843 www.hugeinc.com MARCH 12 TH , 2 01 7 | D E SI G NI NG INTEL LIGEN T CONVERSATIONS @SXSW
- 2. 03.1 2.201 7 DesigningIntelligent Conversations
- 3. AGEN DA 3 1. Intro. 2.Talk. 3. Design. 4. Break. 5. Build.
- 4.
- 5.
- 6.
- 7.
- 8. 8 JesusBot INT RODUCTIONS Helps connectedchristians pray regularly, get advice, and directs donations to great non-profits. 8 @MKLoomis
- 9. Introducing chatbots. DESIGNING INT ELLIGENTCON VER SATIO NS 9
- 10.
- 11. A chatbot issoftware you have a conversation with. DE FININ G CHATBOTS 11
- 12. 12 Anything can be boiled downto a conversation. CHATBOT BA SIC S
- 13.
- 14. 1. Mini-version ofyour business. 2. Extend your marketing. 3. Focus on automation. AWESOME KINDS OF CHATBOTS 14
- 15. 15 Mini-version of yourbusiness. AWESOME KINDS OF CHATBOTS MediaFinance Transportation
- 16. 16 Extend your marketing. AWESOMEKINDS OF CHATBOTS Blog / Notifications Portfolio SiteCampaigns
- 17. 17 Focus on automation. AWESOMEKINDS OF CHATBOTS Customer ServiceOperations Legal
- 18.
- 19.
- 20. 20 Art & scienceutopia?
- 21.
- 22.
- 23. 2323 Guided mediation. Regularreflection. Habitual religiousness AUTOMATION NOW Relationship design.
- 24. 2424 AUTOMATION NOW Is thisreally what we want?
- 25. Time to makesome decisions. 25
- 26. Who do youwant to help? Or replace? WHY THIS MATTER S 26
- 27. What are theroles for bots in the future? WHY THIS MATTER S 27
- 28. What are theroles for humans in the future? WHY THIS MATTER S 28
- 29. Explore what relationshipsdo we want automated by technology & which we don’t. TOD AY’S LESSON P LA N 29
- 30. WHY DESIGN MATTERS Design identifies the right thing to make. 30
- 31. Solve messy problemswith iterative, collaborative & empathetic methods. DE SIGN THINK ING 31
- 32. Time to startdesigning. 32
- 33. Designing conversation. DESIGNING INT ELLIGENTCON VER SATIO NS 33
- 34. Let’s figure outthe goals, structure, & what to say. REME MBER GOOD CO NVER SA TI ON? 34
- 35. Picking goals. DESIGNING INTELLIGENT CON VER SATIO NS 35
- 36. Define the relationship. CHATBOTPERSONA 36
- 37. 1. Parent. 2. Pal. 3.Pet. DEFINING T HE RELATIONSHIP 37
- 38. 1. What doesit tell users to do? 2. What does it do with users? 3. What do users tell it to do? DEFINING T HE RELATIONSHIP 38
- 39. 1. Remind peopleto pray. 2. Help with ethical problems. 3. Send donation to a non-profit. DEFINING T HE RELATIONSHIP 39
- 40. 1. What doesit tell users to do? 2. What does it do with users? 3. What do users tell it to do? DEFINING T HE RELATIONSHIP 40
- 41. Take 5 minto write Parent, Pal, & Pet behaviors. TEAM WORK MAKES THE DR EAM WOR K 41
- 42. What to say. DESIGNINGCONVERSATION 42
- 43. How might wesketch the kind of conversation we are having? THE PROBLEM 43
- 44.
- 45. Using post-it notesto be clear & collaborative. ADVANCED CON VERSATIO N MAPPI NG 45
- 46.
- 47. 47 Clicked buttons Utterances Natural Language Processing(NLP) The script The interaction
- 48.
- 49.
- 50. Pick a Parent,Pal, & Pet goal. Take 10m to sketch out a conversation! TEAM WORK MAKES THE DR EAM WOR K 50
- 51.
- 52. We quickly endup lost in the details. 52
- 53. Stay focused onstructuring interesting kinds of conversations. CON VERSATION STRUCTURES 53
- 54. 54 Try having adifferent kind of conversation. CO NVERSAT ION STRUCTURES String of Pearls Train Track Parallel Track Open World Hub & Spoke
- 55. 5555 CO NVERSAT IONSTRUCTURES Train Tracks for customer service & routing.
- 56. 5656 CO NVERSAT IONSTRUCTURES String of Pearls for news, quizzes, blogs, & notifications.
- 57. 5757 CO NVERSAT IONSTRUCTURES Open World for entertainment & fuzzy searching.
- 58. 5858 CO NVERSAT IONSTRUCTURES Parallel Paths for switching between human & bot, or agents.
- 59. 5959 CO NVERSAT IONSTRUCTURES Hub & Spoke for menus, core stories, & easter eggs.
- 60. Take 5m toidentify the structure of your conversation. Where would you expand to? TEAM WORK MAKES THE DR EAM WOR K 60
- 61. Get ready tomake some bots. DESIGNING INT ELLIGENT CON VER SATIO NS 61
- 62.
- 63. Copy the filesfrom the USB drive to your computer. Super easy. STEP 1 63
- 64. Run the Docker Installer& follow the onscreen instructions. We believe in you. STEP 2 64
- 65. Open the folder named‘demo-bot’. Double click the ‘Install’ file. That’s it for now! STEP 3 65
- 66. Break.6100 Wilshire Blvd,LA CA 90048 www.hugeinc.com
- 67. Making a chatbot. DESIGNINGINT ELLIGENT CON VER SATIO NS 67
- 68. A.I. is notmagic. BUI LD 68
- 69.
- 70. A.I. is “thecapability of a machine to imitate intelligent human behavior.” 70 ― Merriam Webster
- 71. AI is thescience of engineering machines that are capable of non-trivial, non-routine computations. MATT SAYS 71
- 72. AI is asystem that provides the correct service to the user based on the user’s input. DAVID SAYS 72
- 73. AI is anythingthat has been designed with human behaviors in mind. AUSTIN SAYS 73
- 74. The spectrum ofintelligent human behavior. LETS GET C LEAR ON A.I. 74 Very complex systems Made for humans Mapped out actions to reactions
- 75. A.I. IS NOTM AGIC Instead of worrying about A.I., focus on your services. 75
- 76. Machine Learning &API’s help make chatbots feel “smarter.” AI IS N OT M AGIC 76
- 77.
- 78. Machine Learning isa method to detect patterns in large data sets. 78 ― Austin, David & Matt
- 79. 1. NLP /Speech Recognition. 2. Geo-Location and Mapping. 3. Image tagging. EXAMP LES OF MAC HINE LE ARN ING APPLICATIONS 79
- 80. A.I. IS NOTM AGIC Machine Learning is not required for most chatbots. 80
- 81.
- 82. API integrations connectchatbots to services and to the real world. WHY API’S 82
- 83. 1. Getting userdata. 2. Connecting with Internet of Things. 3. Sending a purchase request. EXAMP LES OF API INTEGRATIO NS 83
- 84. A.I. IS NOTM AGIC Creative use of API’s is the real magic. 84
- 85. Now let’s really make thebots. BUI LD 85
- 86. Register at join.riseofthebots.net You willreceive a email confirmation shortly. STEP 1 86
- 87. Go to slack.com/apps/ manage/custom- integrations The linkis in slack. STEP 2 87
- 88. 1. Click ‘Bots’. 2.Click ‘Add Configuration’. 3. Enter a username. 4. Click ‘Add bot integration’. 5. Copy your token. (we’ll use it in a minute). 6. Optional, add a profile pic 88 STEP 3
- 89. 1. Double clickthe LAUNCH file found in the demo-bot folder. 2. When prompted, enter your slack token and press ENTER. 3. Check in Slack to see if your bot is online! 89 STEP 4
- 90. Follow along & getmaking. You are amazing. STEP 5 90
- 91. Rising of thebots. DESIGNING INT ELLIGENT CON VER SATIO NS 91
- 92. Bot imperatives. DESIGNING INTELLIGENT CON VER SATIO NS 92
- 93.
- 94. 1. Quantity. 2. Quality. 3.Relation. 4. Manner. GRI CE’ S MAXIMS 94
- 95. Be helpful &clear. Don’t ramble or have too many options. MAXIM OF QUANTI TY 95
- 96. Be honest aboutwhat you do & don’t over simplify. MAXIM OF QUALI TY 96
- 97. Respond to everything,but remember the goal. MAXIM OF RELATION 97
- 98. Consider how whatis said is said. Be enjoyable, not curt or overwhelming. MAXIM OF MANNER 98
- 99. ALWAYS REME MBER Fuckthe rules. 99 Technically, Grice said “flout the maxims”.
- 100. Done.6100 Wilshire Blvd,LA CA 90048 www.hugeinc.com