Michael Aston, CTO Transversal, gives us some insight into the future of knowledge. Will A.I. take over and destroy us all? or will it make our lives even easier? And where does Transversal fit into the future of knowledge.
10. Natural language interaction
“In the not-too-distant future, users will no longer
have to contend with multiple apps; instead, they will
literally talk to digital personal assistants such as
Apple's Siri, Amazon's Alexa or Google Assistant”
11. What does this mean for Transversal?
We’re putting all our functionality in to APIs to make it externally
accessible – knowledge is a reusable asset.
We’re using context to proactively push relevant information.
We’re tracking behaviour to refine recommendations.
14. › Predictive analytics
The branch of advanced analytics which is used to make
predictions about unknown future events.
Predictive analytics uses many techniques from data
mining, statistics, modelling and machine learning to
analyze current data to make predictions about the future.
15. A quick test
Think of a number between 1 and 10.
On average ~45% of people will choose 7.
16. Patterns in data
* Based on a survey of 34,260 OK Cupid users who
closed their account because they’d found a match
17. › Cognitive systems
A category of technologies that uses
natural language processing and machine learning
to enable people and machines to interact more naturally
to extend and magnify human expertise and cognition.
18. Cognitive knowledge vs. cognitive computing
Transversal Prescience IBM Watson
Answers for common, recurring
questions
Decision support for complex
problems
Delivers immediate value out of the
box
Requires extensive dev. to build a
solution
Minimal training and configuration
required
Months (or years!) of training
required
Curated content Aggregates everything
19. What does this mean for Transversal?
We’re making use of internal and external data to drive
anticipatory interactions.
We’re using cutting edge natural language understanding to
derive intent.
We’re focused on delivering the best outcome for the end user.
21. › Actionable insight – what’s wrong
Monitors my current progress
Alerts me at the exact moment
something goes wrong
Suggests immediate action(s)
to correct my problem
22. › Actionable insight – what’s right
Monitors my current progress
Lets me know how I am doing
against goals or others
Quantifies my progress so I
can earn credit for my
achievement
23. › Actionable Insight – What’s coming
Monitors potential problems
Alerts me when certain criteria
is met, exceeded or abnormal
Suggests possible action(s) to
correct my problem
24. What does this mean for Transversal?
We’re measuring effectiveness and outcomes, not just activity.
We’re proactively identifying knowledge gaps and emerging
information needs.
We’re recommending and prioritizing actions based on impact.
3 trends to discuss on the way to true AI:
Knowledge will be everywhere
Systems will be much smarter at providing accurate answers and anticipating needs
Managing knowledge will be much easier because the system will do the heavy lifting
The future is knowledge everywhere.
Perhaps mention the demise of the app? According to research from Nomura: Snaphat is bucking the trend (source http://www.recode.net/2016/6/8/11883518/app-boom-over-snapchat-uber)
Chat as the new platform
Users are encountering a discovery chanllenge - is there a chat channel available? Facebook Messenger and WhatsApp have the advantage that people know it is there
The Zuck explains messenger bots for us.
Facebook Messenger, Google Assistant, Apple iMessage Apps
http://www.gartner.com/newsroom/id/3352117 - Gartner predicts that, by 2019, in at least 25 percent of households in developed economies, the digital assistants on smartphones and other devices will serve as the primary interface to connected home services. "In the not-too-distant future, users will no longer have to contend with multiple apps; instead, they will literally talk to digital personal assistants such as Apple's Siri, Amazon's Alexa or Google Assistant," said Mark O'Neill, research director at Gartner. "Some of these personal assistants are cloud-based and already beginning to leverage smart machine technology." "APIs are the key to interoperating with new digital interfaces and a well-managed API program is a key success factor for organizations that are interested in reaching consumers in their connected homes," said Mr. O'Neill. "In the emerging programmable home, it is no longer best to spend time and money on developing individual apps. Instead, divert resources to APIs, which are the way to embrace the postapp world."
Current focus is on transactional/business processes systems, e.g. ordering flowers from 1-800-Flowers.com
Self-driving cars - Local Motors 'Olli' - Using IBM Watson IoT to analyse and learning from data from 30 sensors. Also using Watson speech-to-text, natural language classifier, entity extraction and text-to-speech to enable interactions between the vehicle and passengers (how the vehicle works, where they are going and why Olli is making the decisions it is)
Retail Beacons - Apple iBeacon - Bluetooth Low Energy to communicate with devices e.g. iPhone. Microlocation of the device. Activates the beacon - targeted offers, coupon distribution to the device, discounts on leaving for future purchases. Retail, transportation etc.
… but what about providing knowledge in real time to assist in issue resolution? For example combining IoT and AR.
Ask any question any time. Information provided is going to need to be concise. Natural language speech recognition has improved considerably over recent years, up from 75% accuracy a few years ago to 92% today.
http://www.recode.net/2015/3/3/11559650/google-voice-search-talks-me-through-a-house-of-cards-weekend
Amazon Echo - Alexa - play music, answer questions, create to-do lists, shopping lists, set timers/alarms. Can hear from anywhere in the room. Cloud-based. App for control.
Google Home - speaker that plugs into the wall; similar to Amazon Echo; controls home automation devices: thermostats and lights. Exploits Knowledge Graph - entities and relationships - so when a question is asked has context for related questions; makes use of Cast so can display information on the TV.
Possibly include the IBM Financial Services Robot video/Bob Dylan video (see Notebook, Userful Stuff)
The future is knowledge everywhere.
The biggest leap in relevance will come from the various forms of AI and their ability to understand and interpret a person’s intent and link that to information it holds in a knowledgebase.
It seems everyone is doing about AI these days.
AI techniques are being used for everything from playing Go to self-driving cars, creating walking robots and letting computers recognise objects in pictures.
In the world of knowledge, focus will be on two areas:
Predictive analytics
Cognitive computing
Pick a number between 1 and 10
We are bad at being random and statistics reveal those trends.
"It will take six, nine, 18 months to train a domain. It depends on how complicated the domain is, how much data you have," said Chin, adding that training is an ongoing process."
We have case studies across a broad range of industries delivering business value. Proven in contrast to Watson aside from medical applications.
Predicting searches, predicting results
Bypassing journeys where we’re confident of the user action.
Deriving intent, not just keywords.
Understanding how outcome is correlated to behaviour to recognise early signs of negativity.
Knowledge solutions will provide information to let the knowledge managers make efficient, informed decisions.
Technology improvements will help identify when attention is required. E.g. trend watch helps spot user need before it hits your contact centre.
Knowledge systems will be able to identify potential answers for you, by scanning your communities, help desk, etc. to match questions that aren’t being answered to information already held.
Systems will tell you how effectively they’re working and where you the best place to spend your time is to improve them.
Technology improvements will help identify when attention is required. E.g. trend watch helps spot user need before it hits your contact centre.
Knowledge systems will be able to identify potential answers for you, by scanning your communities, help desk, etc. to match questions that aren’t being answered to information already held.
Systems will tell you how effectively they’re working and where you the best place to spend your time is to improve them.
3 trends to discuss on the way to true AI:
Knowledge will be everywhere
Systems will be much smarter at providing accurate answers and anticipating needs
Managing knowledge will be much easier because the system will do the heavy lifting