# AI Expo 2019 ## Convergent Solutions - blockchain, ai, iot ### Daniel Yim - Noblis ### Duane Jacobsen - Blocksafe ### Ian Foley - Xenon (host) ### Kevin Garlan ## Daniel US / Aircraft carriers / consulting (public sector - ) / IBM Focus on data / analytics / viz Eventually blockchain (currently) DC Hyper blckchain association Head of Asia ## Duane CEO Platform as a Service launching stream iot platform (today) giving away 1000 tokens - allows IOT companies physical/hardware/software/application - take data and put it in a blockchain Use cases: firmware caching ## Kevin - north american innovation practice - city treasury banking fortune 500 institutions - glboal network based out of Dublin Questions: Benefits of convergence / what projects ? duane: best way to answer - why is blockchain / iot convergence important? They believe: (coming from financial perspective) financial legering system Take iot devices and secure them Use case: - circuit board can be hashed and have it do something upon awakening - firmware/circuit should not be counterfeit - we need a way to verify this and to verify its usage - Customers develop devices / apps that hash information on the chain / or data off chain / data privatization Follow q: cost saving? increase revenue? a: addresses 2 things: - factors that are coming by 2021 - 75 000 000 000 global devices - devices we carry around (25%) - the rest are "things" -> important that manufacturers can prove privacy and authentication efficacy / inter-device communication daniel: Q: tangible way of increasing revenues / cost changes / How does a CFO view this? a: some level of trial/error in real use cases through fefederal agencies project: - use of blockchain for autonomous vehicles - self driving - future; humans become less able to perform work which machines take over - in this environment, we will have a mixture of human vehicles vs autonomous : how to make this safe? this is closer than we think - Determine how to have machines transact with one another / prioritization - example: UPS' assure that the autonomous driver meets targets -> reqiures priotization: this needs to be paid for - phase 2: other iot devices (roadside devices to verify speed -> share with other vehicles / devices within proximity). compensation to consolidate needs for that prioritization Kevin: Q: how to deploy this tech? Have you done anything a: ths grass isn't always greener. Emerging tech (10 years ago -tart of enterprise practice) - Business lead technology - Human focused design - where can technologies by applied - Best practice: RPA (robotics) - quick sprint from tech to RPA -> infected with broken processes - need a caclculated approach to determining cost/benefit for each tech - BLOCKCHAIN buzzword: is it useful for enterprise business? (it depends?) - currently; starting to see Fortune 500 deploy distrib tech (cars, tomatoes?, airplanes) - Uptake to supply chain finance viewed by CEOs now -> direct cost benefit by removing headcount / digitization Q: How are you doing things differenly from what's being described by Kevin? Duane: a: we think different - our customers are businesses - try to get extra credit on previous question: - ROI for blocksafe is iot token -> utility for transaction - customer -> increase customers via privacy assurance - OTHER LAYER: data privacy. Give businesses a platform to develop whatever they need;. - PIL (person identify info) -> sensor data collection - owning those devices: opportunity for new economy: new people on that device can use platform to aggregate data / agree about how it is used DEFINITIONS Distributed leger Duane; simplest way to loook at this ius think of aleger as an accounting leger: debits / credits. - distributed means having the leger / transactions entered - copied to all nodes follow up question; how is this useful? accounting department is in multiple locations A: enterprise accounting standpoint: instead of having on the cloud, you create immutable transaction seen by every node. Unique solutions identifiers Kevin - everything distributed to identify participants End to End (machine to machine) Daniel: Has a product where this is relevant. Rely on machines to communicate with each other. Security of that information. From machine perspective: increased use cases where information wouldn't necessarily be read by humans anymore What mistakes have you made? Daniel: When bitcoin price peaked - from federal perspective -> the year of white papers - everyone moved contractors around 2018: everyone stand back and see who steps up and does something about this - only a small amount of money spent by government to test this technology - Challenge of where to focus the money on -> techn isn't yet proven - they were weary, btu that landscape is changing. Commercial sector moves faster, of course. Duane: Smaller company - top mistakes would be -> don't go into blockchain thinking it's simple / easy to do. There is no legal knowledge around regulations - this is a floating conversation - grey regulations -> becoming greyer. - Trying to do it all - > this doesn't work. When i became cEO, I pivoted the company from token-based (2016) to company with current platform. Plan is to accomplish the platform at the lower level and work with real companies to demonstrate the usage is legit and purposeful. Machine to Machine payment for priotization - how would this look ? Address safety Daniel: - phase 1 was proof of concept when we started this. We developed a fleet of vehicles that were human controlled / create scenarios where they would crash with one another - > determine inefficiency base don this - phase 2 - vehicles autonomous. Different sensors radaor/lidar/ infrared. Share that info with each other. In environment where they know they are trustworthy - keep full driving record on each machine - > this proves intent. If intent and motion is different, then they reduce their trust score. if we imagine a crossroad where there are no traffic lights - > we wouldnt be able to operate. Machines can digitally communicate when infrastructure is lacking - > tehy are the infrastructure. 'I am okay paying this to do that'. Other machines might say: I will also compensate with x coins, but other machine is only willing to pay x-2 coins. Highest bidder gets the highest priority. Q: understanding the data becomes the data itself. Where is the relationship desirable, but the immutable mess becomes less desirable for said relationship: Duane: example -> our customer is developing distributed app to track firearms. State based rules / laws means behaviour of a firearm within each state needs to change. 5 key components: virearm, accessory, vehicle, origin and destination. AI needs to understan this behaviour for the user. Blockchain creates a private maechanism for the data. application which tracks, tells me, but only i knwo because only I can see the actual decrypted data. Lay persons's answer threat actors; using legacty and analog -> same with blockchain tech -> threat actors acros the world. Not any different compared to cloud rtech / on prem. Up th ante from ca cyber perspective. Bubble to burst - > it's unhackable? not true. Ther eis a lwasys a chance. ------------------------------------------------------------------ IBM Chatbot (Watson assistant) ------------------------------------------------------------------ Automating AI/ML Data Prep enabling contextual intelligence Quadrupling generated data Data Scientist challenges: automating repetitive tasks -> the biggest challenge according research, due to funding and ability to prepare information Current prep mechanism: Unstructured documents, conversations/text/web is the greatest proportion Structuring is usually performed through manual labelling IN perfect world: - organizational unstructured data (docs, notes, agreeeements, contracts, research, marketing, policy/procedure, technical docs, job stuff, articles - articence one click extraction (this product): Industry/domain, entities, file types, relations, label, knowledgegraphs, taxonomy Changed visualization strategy - Why Articence rapid adoption of AI one clickdata ready for ML use low costadvanced graph techbias free domain specific KPIs influenced by Articence - employee satisfaction / retention, etc Rapid adoption of DL/ML -