Autonomous Negotiation What if you owned a personal robot that could negotiate the purchase of your next car? Imagine it could even secure a better deal than you could. What if the same robot could negotiate your apartment rent, next vacation, and mobile phone contract? What if it could even negotiate your salary?
Of course, we’re not quite there yet. Even so, the world of blockchain and IoT continues to make bold strides in creating autonomous negotiating agents. Humans need negotiation classes to improve their skills. However, IoT systems may soon deliver outstanding results right out of the box. There are several ways we can envision today how blockchain and IoT might soon enable autonomous negotiation.
Joint Decision-Making
At the 2017 International Joint Conference on Artificial Intelligence (IJCAI) held in Melbourne, Australia, a team of scientists presented a paper on how Artificial Intelligence could negotiate for humans.
IoT and blockchain technology is already proving to be affordable, reliable, and robust. With the recent announcement of Facebook’s Libra, the blockchain landscape is about to experience even further growth.
IoT devices are already identifying, tagging, tracking, and monitoring our activities. IoT devices are already part of our negotiation support systems.
These support systems are designed to support human decision-making. They are also designed to train people in negotiating. A great example is the Inspire system. This system gathers information during negotiations. It then uses what it learns to recommend outcomes and offer simulation and advice.
Game Theory
In negotiation classes using simulation role plays, game theory is a mathematical-based concept focusing on equilibrium protocols for win-win outcomes. With game theory design, IoT systems can recommend appropriate actions. The response the system suggests depends on the actions of other participants. For instance, imagine the market price of real estate in a certain region goes up. The system would advise the user to adjust an offer for a property they want to buy.
In the current technological environment, IoT devices have limited self-direction. They lack the autonomy to fully act on game theory. As agents, IoT systems can’t make direct bids or adjust strategies. In the future, completely autonomous agents may be able to make judgment calls. They may even be able to fully act on the decision based on the actions of other participants.
Negotiation Analysis
During all negotiation classes, humans learn how to use the responses of others to adjust terms. In negotiation analysis, IoT systems analyze the actions of others to suggest user actions.
Negotiation analysis differs from game theory. Like game theory, negotiation analysis evaluates the actions already committed. However, negotiation analysis also evaluates which actions may be performed as a result of the user’s decision.
A key advantage of negotiation analysis over game theory is that IoT autonomous agents can make decisions without intervention. They can make these decisions with complete authority delegated by the user.
Preference Elicitation
As mentioned, IoT devices have the capacity to monitor, identify, tag, track, and share data about our daily interactions. With increased data connectivity, IoT has a large information database. This database includes information about what you eat, places you go, what you buy, whom you talk to, what you spend your money on, etc.
This data can be used to construct an accurate model of the user’s preferences. By finding out these preferences, autonomous agents can enter discussions on your behalf fully informed. They can then work towards a favorable outcome that will most closely suit your preferences.
Imagine an agent searching for a used car based on your daily mileage, budget, and even type of commute. The agent could then negotiate the best possible price before presenting the deal to you. What’s more, the agent could even execute this routine many times with different dealerships and present you with a list of cars to choose from.
The current challenge for preference elicitation surrounds privacy issues. Many users are also reluctant to engage with IoT systems at length.
Some top tech companies have faced allegations of security and privacy abuse. As such, it may be a while before users are ready to share the personal data required for effective negotiation.
Domain Modeling
The negotiation outcomes provided by IoT and blockchain systems depend on preference models. They also depend on the system’s accuracy in creating a user’s domain model.
Domain modeling is the total knowledge about the user (domain) and an abstraction of the user’s logic (modeling).
Most FinTech transaction scripts use a basic form of domain modeling. This is where an action results in a series of processes and procedures. With the extent of IoT domain modeling, we are likely to see autonomous agents engaging in more complex negotiations.
For instance, to book your next vacation and set in motion savings for your vacation, IoT can use factors like:
- Past preferences
- Financial position
- Family preferences
- Accommodation
- Transport availability
- Weather patterns
Nonstationary Preferences
Humans have varying opinions and preferences over their lifetimes. IoT can not only track your behavior over time but also note changes in behavior and needs. As the user’s preferences evolve, so does the value attached to particular things.
The challenge is if the IoT system fails to adjust to nonstationary preferences, the user may notice a drop in performance. A couple expecting a baby may vary their preferences compared to when they were childless. If the system doesn’t update preferences, the user may then be wary of continuing to use autonomous negotiation agents.
A simple example is when buying Christmas gifts for your son. If last Christmas he was six years old, his preferred toys were probably different than what he wants now. The autonomous agent should attach a lower value to last year’s toy preferences. A higher value should be applied to the prevalent tastes.
Algorithms and AI in Teaching Negotiation
One major challenge has plagued the negotiation classes field. How can one facilitator provide insightful actionable feedback to a classroom of participants? Only one pair of ears and eyes divided across perhaps tens of teams of course participants.
With each individual in each of these teams having their own needs to be met, it’s easy to see how classroom teaching can have its limitations.
The Negotiation Experts created a solution to this age-old learning bottleneck. Their gamified negotiation simulation provides instant feedback to users. This feedback takes the form of points scored.
Graphs at the end of the sim show where the player created, exchanged, and destroyed value. So measuring the degree to which negotiators have created a true win-win solution (or win-lose) is now possible. The next step is to move from smart algorithms to AI.
The Role of Blockchain in Autonomous Negotiation
Blockchain technology promises the world a global market free of manipulation. Blockchain provides accurate identification through digital certificates, a medium of exchange, smart contracts, and reliable bookkeeping.
While smart contracts may not yet replace lawyers, they are reliable enough to enter into, enforce, and implement simple agreements. Blockchain bookkeeping can perform instant audits. It can also run ledger analytics to optimize the system.
IoT and blockchain technologies can be integrated to create autonomous negotiation agents. The system can work to make joint decisions with favorable outcomes for the user depending on preferences and domain modeling.
For instance, imagine your IoT system at home making a deal on your annual gym subscription. Depending on factors like how often you exercise, how much you can afford, and what time you exercise, the IoT system makes a deal on an extra-discounted Silver package subscription for your exercise goals.
The future is fast arriving. We need only tackle the controversial areas, especially the inherent privacy and security concerns.
Article Abstract
Blockchain and IoT technologies can be integrated into joint decision-making devices. With negotiation strategies backed by domain modeling, autonomous negotiation agents can work to provide close-to-perfect outcomes for many interactive situations.
