This document summarizes a presentation on blockchain for multi-agent systems regarding trust, reputation, and ethics. It discusses how blockchain can be used to manage trust and reputation in multi-agent systems. It presents an example case study of a startup assessment platform using blockchain and agents. It also outlines some open technological and ethical challenges around integrating blockchain and multi-agent systems, such as scalability, standardization, privacy, and liability.
1 of 31
Download to read offline
More Related Content
BCT4MAS - Blockchain for Multi-Agent Systems - NTN invited talk
2. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
2Introduction
Davides Research Areas
Real-Time Multi-Agent Systems (RT-MAS)
Blockchain & Multi-Agent Systems (BCT4MAS)
{B0}
{B1}
{B2}
{E1}
{E2}
Proximal joint
Distal joint
Distallink
Proximallink
Digital (tele)rehabilitation (DTR)
Explainability in MAS (XMAS)
Real-Time Tire Monitoring (SmartTyre)
Ambient
Assisted Living
Safety-critical
Applications
Tourism
Distributed AI
eHealth
3. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
3
hard-coded, semi-automated, autonomous
static, uncertain, unpredictable
Scenarios
Introduction
Strict Dependability
Timing Guarantees
Embedded Systems
Safety and Security
Scarcity of Resources
Abstraction
Accountability
Trusted entities
Trusted data
Trusted interactions
Distribute Thinking
Flexibility
Recon鍖gurability
Robustness
Ubiquity
many systems, still one challenge
4. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
4MAS & BCT
Image: Peter Bergstrom
5. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
束An agent is anything that can be viewed as perceiving its environment through sensors
and acting upon that environment through effectors.損
5
[Russell & Norvig]
束Autonomous agents are computational systems that inhabit some complex dynamic environment, sense and act autonomously in this
environment, and by doing so realize a set of goals or tasks for which they are designed.損 [Maes]
束...software components that communicate with their peers by exchanging messages in an expressive agent communication language.Agents can be as
simple as subroutines; but typically they are larger entities with some sort of persistent control.損 [Genesereth & Ketchpel]
束We consider agents to be systems that are situated in some environment. By this, we mean that agents are capable of sensing their
environment (via sensors), and have a repertoire of possible actions that they can perform (via effectors or actuators) in order to modify
their environment.損 [Wooldridge]
Multi-Agent Systems in a nutshell
6. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
6Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
Independent decision making
(without human action)
7. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
7Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
To execute
GOAL driven actions
8. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
8Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
Migration on
several network nodes
9. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
9Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
Self-adaption to
user needs, environmental needs
10. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
10Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
Communicate with other agent
to (share - require - contract)
informations - performance
11. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
11Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
Always GOALs oriented
Never to avoid
their achievement
12. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
12Multi-Agent Systems in a nutshell
Knowledge
Behavior(s)
Interaction Policies
Autonomy
Social Ability
Reactivity Proactivity
MobilityRationality
Machines Learning
React to environmental changes
or due some events
13. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
13Multi-Agent Systems in a nutshell
Reasoning upon Explicit
symbolic representation
Logic is easily understood
Robustness
Response Time
Complexity
faithfully world
representation
Require High
standardization
Security & Trust
14. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
14MAS Design
Domain
Model
Computational
Persuasive
User
Pro鍖le
State
Machine
Coordinator
Mediator
Coach
privacy sphere
Behavior change Agent
Calbimonte, J. P., Calvaresi, D., Dubosson, F., & Schumacher, M. (2019,
June). Towards profile and domain modelling in agent-based applications
for behavior change. In International Conference on Practical Applications
of Agents and Multi-Agent Systems (pp. 16-28). Springer, Cham.Behavior
15. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
15Trust in MAS
a belief an agent has that the other party will do what it says it will
(being honest and reliable)
or reciprocate
(being reciprocative for the common good of both),
given an opportunity to defect to get higher payoffs
Calvaresi, D., Calbimonte, J. P., Dubovitskaya, A., Mattioli, V., Piguet, J. G., & Schumacher, M. (2019). The good, the bad, and the ethical implications of bridging blockchain and multi-agent systems. Information, 10(12), 363.
16. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
16Introduction
BCT in a nutshell
Scope Whats a block chain?
Ledger
Distributed
Secured
Immutable
Transaction 20
Transaction 21
21/6/18 12:00
038314966
0b53a3166
Transaction 22
Transaction 23
Transaction 24
21/6/18 13:00
0b53a3166
110c5321a1
Transaction 25
Transaction 26
21/6/18 16:00
110c5321a1
xx1c5x3a1
Block30 Block31 Block32
Block ID (hash of the contents)
Transactions
TimeStamp
Previous Block ID
Logic
Replicate the single DB
and the logic
17. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
17Introduction
BCT in a nutshell
LEDGER
WORLD STATE
UPDATED
LOGIC
1
2
3
4
5
6
Transaction (TXs) sent to the Peer
Peer invokes Smart Contract
Peer uses WS to generate the TXs
TXs grouped in a new Block
New Block added to the BC
18. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
18Introduction
Possible scenarios
BC-A*
BC-A*
19. 19Certi鍖cates and Identity management
| |Davide Calvaresi Trusted Reg, Neg and Service eval in MAS via BCT WI, Santiago - 4.12.2018
Membership
Service
AMS
DF
Can I
Join?
Here is
your eCert
1
2
3
5
Let me
see
(CA-A)
WHO WHAT
ag_1 serv_1
ag_1 serv_2
ag_2 serv_1
4
(BC-A)
1 2
3 4
5
20. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
20
BC-A
InvokeChaincode
Assets Ledger
Register Agent
Add Service
Register Interaction
Map Agent-Service
Edit Agent
Edit Service
Delete Agent
Delete Servie
Create Activity
Update Reputation
CA-A
Membership Service Issue Valid Identities
Identify Roles
Manage CRL
Certification
Authorities
Core BehaviorView Controller Model
Register Agent
Add Service
Add Interaction
Edit Agent
Edit Service
Delete Agent
Delete Service
Update Reputation
Issue Certificate
Verify Credentials
Current Architecture
21. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
21
IdService
Name
Description
Cost
Time
...
Service
IdReputation
IdService
IdAgent
AgentRole
Value
Reputation
0..n1
IdServiceAgentAssociation
IdService
IdAgent
Cost
Time
ServiceAgentAssociation
0..n
1
IdAgent
Name
Address
...
Agent
0..n 1
IdInteraction
IdWritierAgent
IdDemanderAgent
IdExecutorAgent
IdExecutedService
TimeStamp
Value
Interaction
0..n
1
0..n
1
0..n
1
Current Architecture
indexes
22. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
22
|
Case Study
InnMind Startup assessment platform
Startups InvestorsExperts
Startup Expert Investor
Startup B0, B11 B2, B5, B10, B11 B3, B11
Expert B5, B10, B11,B12 B12 B0, B4* B6, B11
Investor B8, B11 B9, B10, B11 B0, B7
maximizing their interests
having freedom of joining, serving,
and leaving the platform at any time.
organize coalitions
foster sel鍖sh interests manipulating and
exploiting other actors or some dynamics in the platform
Behavior Action
B0 Actor pro鍖ling
B1 Startup self-assessment
B2 Request of assessment
B3 Request visualization
B4 Exner assessment
B5 Assessment response
B6 Suggestion
B7 Success rate
B8 Demand for an expert
assessmentB9 Demand for assessing a startup
B10 Assessment negotiation
B11 Startup evaluation
B12 Assessment proposal
Calvaresi, D., Voronova, E., Calbimonte, J. P.,
Mattioli, V., & Schumacher, M. (2019, June). A
startup assessment approach based on multi-agent
and blockchain technologies. InInternational
Conference on Practical Applications of Agents and
Multi-Agent Systems(pp. 67-79). Springer, Cham.
23. Blockchain for Multi-Agent Systems: Trust, Reputation, Ethics| |Davide Calvaresi TNT group 17.9.2020
Case Study
InnMind Startup assessment platform
EXP2STUP1 EXP1STUP2
(a)
(b)
(c)
B0
B1
B0 B0
B2
(B10)
(B10)
B5
(B11)
(B10)
B5
B11
B12
B0
B2
23
24. Blockchain for Multi-Agent Systems: Trust, Reputation, Ethics| |Davide Calvaresi TNT group 17.9.2020
Case Study
InnMind Startup assessment platform
eval
BCT EX DM
Service Negotiation
donedone + eval
ok - revise
revise eval?
(a)
EX = Executor; DM = Demander
no
revise eval?
(b)
Disagreement resolution policy
eval(EX)
eval(DM)
24
26. Blockchain for Multi-Agent Systems: Trust, Reputation, Ethics| |Davide Calvaresi TNT group 17.9.2020
Case Study
InnMind Startup assessment platform
26
27. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
27Discussion
Lesson learnt in BCT 4 MAS
WHEN and HOW to apply
in function of requirements and context.
BREAKTHROUGH and DOMAINS
(DIS)ADVANTAGES
OLD or NEW
technologies/techniques?
TECHNOLOGIES
Prevention of Exploitation and Deception
Spot (un)intentional biases
Remove the need for TTP
Non-negligible Overhead
Complex integration
28. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
28Discussion
TECHNOLOGICAL Open Challenges of BCT 4 MAS
Multiple (Membership Service + Orderer) entities in permissioned BCT;
Scalability in coupling BCT and MAS;
Standardization and Reconciliation of SC for reputation mgm in holistic systems
Ensuring reliability of the mechanisms bridging the two technologies
Verify correctness of (new/amended) smart contracts
Improve privacy and anonymity
Technical immutability of smart contracts in enterprise software contexts
Can an AI entities (i.e., agents) autonomously develop SC?
29. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
29Discussion
ETHICAL Open Challenges of BCT 4 MAS (i)
Erosion of the socio-cultural blockchain ecosystem
(technological accessibility and egalitarian ecosystem?)
Smart contract complexity as Blockchain Fraud
(unexpected hard-to-foresee emergent properties and ML?)
Theoretical and effective empowerment: from accessibility to workability
(workarounds and implications?)
The persistent need for intermediaries to implement smart contracts
(who and how develops SC and liability?)
Calvaresi, D., Piguet, J. G., Calbimonte, J. P., Kampik, T., Najjar, A., Gadek, G., & Schumacher, M. (2020, October). Ethical Concerns and Opportunities in Binding
Intelligent Systems and Blockchain Technology. In International Conference on Practical Applications of Agents and Multi-Agent Systems (pp. 5-16). Springer, Cham.
30. BCT4MAS: Trust, Reputation, Ethics| |Davide Calvaresi ZHAW, NTN group 17.9.20
30Discussion
ETHICAL Open Challenges of BCT 4 MAS (ii)
Liability in AI delegation
(in complex systems, who is contracting, who & why should be liable for that?)
Constructing a fair reputation-building system through BCT and MAS
(authenticity, origin, traceability, and accuracy?)
Right to be forgotten
(old but gold)
Create a legal base for BCT and AI
(from where should we start?)
Calvaresi, D., Piguet, J. G., Calbimonte, J. P., Kampik, T., Najjar, A., Gadek, G., & Schumacher, M. (2020, October). Ethical Concerns and Opportunities in Binding
Intelligent Systems and Blockchain Technology. In International Conference on Practical Applications of Agents and Multi-Agent Systems (pp. 5-16). Springer, Cham.