AIGC Token Whitepaper

21 min readApr 9, 2018

Abstract

AI Gaming is an existing artificial intelligence education, competition, crowdsourcing, branding, recruitment and software collaboration platform driven by the AIGC cryptocurrency token. We propose to crowdsale a quantity of our token to fund the expansion of the platform by adding new users, new customers, building on our existing educational offering and further utilising blockchain technology with the goal of becoming the number one global resource for AI development expertise.

Introduction

Artificial intelligence (AI) is rapidly emerging as a transformative technology. Recent advances in the availability of computing power and large training data sets have allowed machine learning algorithms to tackle problems of impressive complexity. From mastering the game of “Go” to powering voice-controlled assistants, AI is already making an impact. Yet we only stand at the threshold of a revolution.

The global benefits of AI are estimated to exceed $15 trillion by 2030. An AI industry is scaling up to deliver these benefits, with a predicted annual growth rate of 52% over the next eight years. The AI industry alone is expected to generate annual revenue of $58 billion by 2025. However, there are some constraints that are currently holding back the development and wide distribution of the benefits of AI.

Primarily, there is a massive shortage of AI developers. As we move into an AI-driven future, the demand for these skills will only increase. Coding for AI requires a level of technological expertise and complexity that demands a highly specialised skill set. Secondly, the industry suffers from a silo effect where the largest tech companies strive to monopolise AI talent as they compete for dominance. Finally, a crisis of replication is keeping some AI breakthroughs stuck in the lab. Without access to the full algorithm, training data or development environment other researchers frequently cannot reproduce the same excellent results. AI Gaming tackles all three constraints.

Our competitive bot vs bot platform objectively identifies AI expertise and makes it available to all organisations, irrespective of size or location. We will connect employers and users of AI services to proven AI expertise and algorithms from around the world. By drawing in and educating the next generation of AI developers, we actively grow the talent pool. Our mission is to be the number one global resource for AI outsourcing and recruitment.

This goal is made possible by the use of a native, decentralised currency token, the AI Gaming Coin or AIGC. We leverage the power of the blockchain to immutably store users’ performance and history on the site, as well as to incentivise them with financial rewards. Continual competition drives a form of natural selection among bots as our developers strive to upgrade and improve their code to find optimal solutions to each of our gamified real-world problems.

AI Gaming Overview

Using a native cryptocurrency (the AIGC) enables AI Gaming to provide a powerful combination of features for the development and distribution of AI.

Education

We offer live and online courses where our users can earn certificates of proficiency. Users are paid in AIGC to complete our courses. We call this “earn as you learn”.

Competition

Users compete and demonstrate their abilities in continual games, challenges and tournaments. The tokens are their entry fee, a reward for success and proof of proficiency.

Crowdsourcing

Customers that have problems that may be solved by AI can outsource to our pool of certified AI developers. Crowdsourcing customers make AIGC micropayments to all users who participate in solving their problems.

Recruitment

Employers and agencies can pay AIGC to access our recruitment system, with the proceeds shared with the users that have opted into it. Recruitment customers can view a form of virtual CV of AI proficiency which details past performance at an objective, granular level. We plan to migrate this to a non-falsifiable decentralised record which employers can be confident reflects reality.

Branding

Organisations keen to associate with the AI revolution can request the development of white label games to carry their name. Such games run on our platform, powered by AIGC, accessed from their website or by organic users. Educators can also have our AI certificates branded for their own use.

Collaboration

Our users are incentivised with AIGC to create specialised AI agents, or “bots” that offer services for other bots to employ via our Bot Service Marketplace. AIGC micropayments are made for services rendered, encouraging collaboration.

Why Use Our Own Blockchain Token?

A blockchain platform will be used to immutably store users’ educational progress, competition results and crowdsourcing history and feedback. When a user chooses to opt into our recruitment system, crowdsourcing and recruitment customers can browse this history to determine with whom to interact.

Using our own cryptocurrency also means that the market forces of supply and demand operate to encourage more participation by developers when there is a shortage of developers or an excess of demand for AI services, and vice-versa.

With this ecosystem, we can provide bespoke AI services to all organisations of any size worldwide in a scalable, accessible manner, democratising both the supply and utilisation of this exciting new technology.

History and Traction of AI Gaming

Founded in 2016 in Oxford, United Kingdom, AI Gaming has developed an innovative approach to build a community of AI developers sourced from both the top universities around the world and the AI community at large. On our platform AI developers create computer programs to solve various challenges, then test them in competition with their peers.

AI Gaming has already passed several milestones:

Over 1.1m games completed by over 7,000 registered users.
Successfully completed 6 hackathon events with sponsorship from Bloomberg, TPP and RS Components amongst others.
Secured Microsoft to sponsor the next AI Gaming hackathon event, to be held at Oxford University Mathematical Institute. Students will be invited to compete to solve a Microsoft branded challenge utilising their Cognitive Services APIs.
Presented a series of lectures at Exeter College, Oxford University. These will form the basis of AI Gaming’s first certificate and MOOC.
Reached the second round of the $5 million IBM Watson AI XPRIZE, one of only three UK teams remaining.
Achieved a Facebook post reach of 200,000 per month, with 8,000 followers.
Launched nine challenges online, with 3 more currently in development on our test platform.

Outline of Vision & Core Objectives

Education of the next generation of AI developers
Crowdsourcing of AI solutions to real-world problems
Democratising access to customers of AI services
Helping employers and candidates in machine learning and AI to connect
Assisting companies to market their products to machine learning enthusiast.
Hosting a competitive AI vs AI arena to accelerate bots’ evolution at specific tasks
Establishing an environment where AI performance can be reliably measure.

Market Overview: The Artificial Intelligence Industry

The world of Artificial Intelligence got its start in the academic realm, and for decades has remained confined there. In the 21st century, AI has gained traction beyond academia, predicated by a combination of economic factors and bolstered by a new era of advanced technologies including:

Increased machine computational capacity via improved hardware capabilities (CPU, GPU, ASIC, and FPGA), storage capacity, and parallel processing;
Refinement of machine learning and deep learning algorithms;
A massive influx of available data via Web 2.0 and, eventually, Web 3.0 which greatly influenced the rise of Big Data.

The intense media and marketing attention surrounding AI is often driven by the misunderstanding as to the true complexity of creating and deploying “thinking machines.” But, this is overshadowed by the industrial push towards implementing AI to create self-driving cars, constructing “smart” cities, detecting and finding solutions for potential fraud, performing autonomous functions in jobs that are dangerous to humans, and learning our emotional or physiological patterns for early diagnosis of life-threatening symptoms.

Clearly, the AI market will continue to expand at a quick pace over the forthcoming years. This growth will lead to an increased demand for AI experts as a wide variety of industries begin to leverage the technology to improve existing automated processes or implement new AI systems that fully automate both basic and more complex tasks that, in prior history, only human labour was able to perform.

Worldwide AI Revenues 2016 to 2025 (In USD Billions)

AI Industry Key Market Drivers

Consistent evolution of new and advanced artificial intelligence.
Rising adoption of artificial intelligence across different verticals of end-use industries.
Heavy investments towards the development and applications for artificial intelligence by industry giants.
Algorithms and access to tools: algorithms like deep learning and its hierarchical pattern recognition are stepping in as a major driver of AI adoption.
The progress of new innovative AI startup companies is constrained by the availability of human talent. Companies compete for this scarce resource.
Artificial intelligence can transform the productivity and GDP potential of the global economy. Strategic investment in different types of AI technology is needed to make that happen.
Labour productivity improvements will drive initial GDP gains, as firms seek to “augment” the productivity of their labour force with AI technologies and to automate some tasks and roles.
PWC research shows that 45% of total economic gains by 2030 will come from product enhancements, stimulating consumer demand. AI will drive greater product variety, with increased personalisation, attractiveness and affordability.
Intellectual Property: Machine learning patents grew at a 34% Compound Annual Growth Rate (CAGR) between 2013 and 2017, the third-fastest growing category of all patents granted.

AI Employment Market

The demand for AI talent is fierce, and academia is often the target of technological behemoths, such as Google, who “poach” AI developers and professors from Russell Group and Ivy League universities, and other high ranking schools. Lured by six-figure salaries, with some companies offering between $300,000 and $500,000 annually, it’s little wonder that the likes of Baidu, Amazon, and Microsoft are successful in enticing academics to leave their lofty university positions. However, according to The Organisation for Economic Co-Operation and Development (OECD) small to medium-sized businesses (SMEs) comprise over 95% of the total number of enterprises that represent the 35 nations who are members of the organisation. Furthermore, SMEs employ 60% to 70% of the working population within those same nations. AI Gaming’s platform provides these companies with direct, scalable access to accredited AI talent.

The AI Gaming Solution

AI Gaming aims to build the world’s largest community of AI developers via its education and gaming platform (https://aigaming.com/). Developers are incentivised to learn and exercise AI development skills by receiving proportionate rewards of cryptocurrency.

Education

High demand for AI developers is driving a global machine learning courses market, projected to be worth $8.7 billion by 2021. AI Gaming’s unique structure means it is well placed to capture a significant slice of this fast-growing market.

To build the broadest user base students are rewarded in AIGC for taking the introductory AI Gaming certificates. For each step a student completes, the site rewards them with a small amount of our token. Measures of financial value on the aigaming.com platform are stated as mA which is short for milli-AIGC, 1/1000th of an AIGC token. Their submitted code and results will be stored in the blockchain so that their progress will be available in an immutable form for future review by recruitment agencies or so users can prove their competence to crowdsourcing customers.

For example, a particular step in a certificate could be to use a technique that they have recently learned to play against a house bot (a bot written by AI Gaming to provide a fixed standard of competition). Their results, including what percentile their results are versus their peers, would be stored on the blockchain.

We are developing a range of certificates which will enable users of all proficiencies to demonstrate and prove their skill set. Through top-level certification and dominance in competition, AI Gaming offers a showcase for developers up to the highest level to achieve peer recognition and career advancement.

Additional certification levels being developed include:

Adding Intelligence
Using Probability
How to use the numpy library in Python
Implementing Microsoft Cognitive Services APIs
Bonus Certificates such as Complete Recruitment Profile

All certificates will either:

Promote engagement with the website
Teach and certify competence in AI programming languages and concepts
Teach and certify competence in AI tools and technologies — these certificates may be sponsored by a branding customer.

In the second and third categories, passed certificates are available to users to add to their LinkedIn profiles, with AI Gaming being the certifying authority.

We can create bespoke certificates which may be branded by customers or educational institutions to their requirements. We believe this will be a convenient way for educators to add an AI component to an existing course, or motivate students to take AI modules by introducing gameplay to AI studies. We offer a way to grade individual students’ progress and benchmark their department’s achievements to an international standard of AI proficiency.

We believe AI Gaming’s reward-based education system is well positioned to help fill the existing gap between the industry demand for excellent AI developers and their supply.

Certification Example

Anne joins the site and opts to take the Certificate 0 course immediately. The AI Gaming Online Code Editor (OCE), will guide her through the process of achieving the certificate. Certificate 0 is simply a way to familiarise a new user with the system.

For example, the first three steps of Certificate 0 are:

Play a game of Noughts and Crosses (1000 mA reward)
Set the Thinking time (500 mA reward)
Understanding how error messages are displayed (500 mA reward)

In addition, when Anne completes all 17 steps in Certificate 0, she receives a certificate completion bonus of 10,000 mA.

Competitions and Tournaments

AI Gaming runs physical and virtual hackathons where users can play challenges in a fun competitive environment, win prizes, and meet interesting sponsors. Prizes will include amounts of AIGC. We also run regular competitions on our site where users can win AIGC and other prizes, in addition, we host continual leagues for each game we offer. At the head of the site is a leaderboard of the best bots across all game types.

Users can enter challenges ranging from two-player games to multiplayer tournaments and use AIGC as their stake and receive AIGC as their reward. Challenges range from popular games such as battleships to real-world problems which introduce users to skills they would use on our crowdsourcing platform.

AI Gaming is not a gambling site. Only skill games and challenges are supported on our platform. By simply having user created bots play for small amounts of cryptocurrency versus using worthless points, we notice a marked improvement in code quality as users take the challenges far more seriously.

The results of a user’s competitions and challenges will be stored in the blockchain and will be available for crowdsourcing or recruitment customers to review.

Crowdsourcing

Users who have reached a certain level of certification can be invited to participate in the crowdsourcing system. AI Gaming works with customers to gamify their real-world problems and offer them to the bot network as a challenge. Users play the challenge competitively among themselves for small stakes of AIGC, inherently encouraging them to improve their bot’s performance from the outset.

Customers can then send queries on an ongoing basis to an API provided by AI Gaming and the code written by the users run to respond with their results. These results will be processed and presented to the customer, who will be able to respond with feedback on the quality of the results if they wish.

Numerous bots may work on a particular customer’s crowdsourced problem, providing a “wisdom of the crowd” solution. We expect novel approaches, which may not be available using traditional outsourced development, to appear as a direct result of our crowdsourcing approach.

Details of bot participation and customer feedback will be written to the blockchain. Where bot code is run on the AI Gaming platform (as would be the case for confidential data) the signature of the code will also be written to the blockchain, allowing future customers to confirm that the same code was used while still protecting users’ IP. Because bots are developed and evolve within our controlled environment, results are consistent and can be repeated reliably in the future. Users will be rewarded for crowdsourcing participation with AIGC which is ultimately paid by the customer.

Crowdsourcing customers have access to a browser that shows them the user’s educational, competition and crowdsourcing history, to help them to decide which users they allow to work on their problem. The customer has the option to select which individual bots can respond, based on such factors as experience or feedback.

We have identified many areas where crowdsourcing of AI development would be beneficial and have had initial discussions with several customers. Typical examples fall into the categories of optimisation, prediction and recognition.

Explored examples include:

Fresh food sales level prediction in a grocery store chain
Recognition of number plates from poor quality CCTV images to build witness lists to assist in police investigations
Identification of scalpers’ (touts) bots purchasing concert tickets online
Logistics analysis to optimise stock picking in a retail warehouse

Number Plate Recognition Example

The customer has a requirement to identify the correct number plate given a poor quality image and a list of potential values. An example might be a police force wanting to build a witness list where the only available imagery is CCTV footage from stores adjacent to a road. The images may not be human-readable; they may be partially obscured or out of focus. The list of potential number plates may be the entirety of vehicles on the road or a much smaller list, for example, one built from a bridge camera near the point the image was captured:

We work with the customer to gamify the problem of recognising number plates. This involves the creation of a challenge on our platform that closely replicates the required skill set for the real world challenge. In the absence of labeled data, we create synthetic data. Bot development undergoes three phases:

Synthetic labeled test data is produced and made available for download by interested users. This allows them to develop their code offline of the system.
The gamified challenge is made available on our system for inter-bot competition. A launch competition is held with a prize to publicise the challenge and attract a greater numbers of users. Users play each other in continual competition with a financial incentive on every game.
Once the bots have demonstrated a level of performance useful to the customer (for example the ability to identify number plates that are not recognisable to humans), the services of the bots are made available to the customer via an API. The customer may feedback on results and at his option, use the accuracy of bots’ previous performance to specify which may handle future requests.

Recruitment

Users can opt into a recruitment system and choose to show their educational, competition and crowdsourcing performance to employers or recruitment agencies.

Recruiters will pay small amounts of AIGC to users to have access to their data as part of the recruitment process. Recruitment customers can view a form of virtual CV or resume, a document similar to an academic transcript which lists the user’s educational progress, competition history and results as will be recorded on the blockchain.

Employers can browse our database of workers from around the globe and view their performance in standardised tests. For example, if their ideal candidate was located in Brazil or India, a traditional recruitment path might never have found them.

Challenges can be custom developed to suit a particular company’s recruitment needs. For example, a company that required developers to work on fresh food level sales prediction could create a challenge which exercised skills of time series prediction and regression analysis and provide authentic historic test data for machine learning. They could compare results from many participants to score candidates on their proficiency at the very competencies they will require in their positions. Employers can even examine the code submitted by participating users for factors such as style, quality and efficiency.

Bot Collaboration

Bots may use services supplied by other bots. Specialised bots can offer their services via an online system — the Bot Service Marketplace, where developers can browse and identify opportunities to contract with third party bots that excels at particular functions.

So for example, in the case of number plate recognition, a first step may be to identify where in an image of an entire car the number plate is located. Rather than implement this in the bot directly, the developer might choose to contract the services of another bot which specialises in detecting the coordinates of the number plate part of an image of a car. The primary bot which has been contracted by the crowdsourcing customer to solve the problem would then pay some part of its compensation to the bot offering that service.

The AI Gaming platform handles the transmission of data between the bots and the subsequent transfer of AIGC balance as payment for the service provided. This feature enables bot with bot cooperation to solve complex problems and offers developers an opportunity to earn from components of solutions they have developed and made available to others, encouraging specialisation and collaboration.

Bots may also call third-party services such as those offered by Microsoft Cognitive Services, and we will investigate integrating with other emerging AI service marketplaces such as SingularityNET, as they evolve.

Summary of the AI Gaming Solution

Notably, the AI Gaming blockchain will not only be a decentralised store of user balances but also of bot performance. We provide a fun environment for playing games, drawing developers into coding for AI, while improving AI development skills. Games are played for small stakes of our native cryptocurrency token AIGC. We offer online courses and accreditation to users which they can put on their CV or LinkedIn profiles.

Qualified users who have achieved a certain level of certification can participate in our crowdsourcing system where they will earn AIGC. We also offer a convenient, opt-in mechanism for them to connect with employers and demonstrate their skills and coding techniques.

Using a cryptocurrency democratises the AI development industry by allowing programmers from around the world to be rewarded directly for their performance, irrespective of location, language, or even prior qualifications. At AI Gaming, only results matter.

The irreversibility and transparency of records stored in a blockchain mean that the records stored are permanent, chronologically ordered and available to all others on the network, thus ensuring integrity.

A set of open standards will allow other platforms similar to aigaming.com to be created by third parties.

The AI Gaming Platform

The AI Gaming platform (http://aigaming.com/) currently stores player balances and results in a centralised database. The adoption of an Ethereum ERC20 token, and ultimately an alternative blockchain platform with higher transaction capacity, follows this roadmap:

Phase I (complete)

Users’ bitcoin balances are converted to Ethereum ERC20 token: AIGC
Users have the ability to withdraw AIGC to their own wallets
Users’ balances and results remain stored centrally in our database

After Phase I and listing on an exchange, we have the benefit that supply and demand market forces find the utility value of AIGC. A shortage of developers would increase the price of AIGC, as crowdsourcing customers need to buy more token to attract them to their problems. The increased price would attract more developers to spend their time in crowdsourcing challenges and to participate in educational certificates, increasing the supply of developers. A reduced AIGC price will make our platform more attractive to customers, leading to increased demand.

Phase II

Identification of an alternative blockchain, which can support the transaction volume and data storage requirements of the platform
Storage of user balances directly on the blockchain
Storage of user performance directly on the blockchain

After Phase II, users have the certainty that their balances are safe in their own decentralised wallets, and customers have the certainty that the performance metrics that they see on the blockchain are true and can be independently verified if necessary.

Phase III

Create a set of open standards to allow third party sites to write to and utilise the information in the blockchain. This would enable an alternative site like aigaming.com to be created, stimulating healthy competition and allowing market forces to better determine the value of the platforms.
Integrate with third-party AI bot marketplaces such as SingularityNET and others that emerge, so that AI solutions may draw on an even wider field of expertise.

Use of Funds From Token Sale

Recruit More Users

Our ambition is to become the world’s primary resource of crowdsourced AI solutions and AI recruitment. To that end, we intend to use funds raised to make developers aware of our platform and its benefits. This will take the form of traditional marketing, as well as outreach at live hackathons and conferences.

Expand Educational Programmes

Funds raised during the token sale will be used to expand our university outreach by orders of magnitude. By employing ambassadors to represent us at university hackathons and conferences, and by incentivising university societies to use our platform, we will be able to make aigaming.com a staple resource for students interested in AI throughout the United Kingdom and internationally.

We are developing a “course in a box” which includes presentation slides, lecturer course notes, YouTube videos, GitHub source code repositories and Google Classroom resources. The format was piloted with the Oxford University Coding Society during the second term of academic year 2017/2018. With the funds gained from the token sale, we intend to trial the course at ten further university societies in the following term, and then offer it to all relevant UK university societies in the first term of academic year 2018/2019. We will then roll our program out internationally, alongside the development of more advanced courses.

Accelerate Product Roadmap

Raising funds via a token sale will enable us to accelerate our development roadmap. While the elements of our system are in place, we will continue to enhance and improve our current functionality to make it more accessible and scalable, and introduce new features such as continuous bot play and inter-university leagues and competitions.

Marketing to Crowdsourcing and Recruitment Customers

Our ecosystem benefits from having a wide range of customers for AI development services. We intend to use some of the funds raised to raise awareness of our system’s benefits to employers and agencies in these customer segments.

It will benefit our users to have the greatest possible range of employers and recruitment agencies represented on the site so that they have the broadest possible choice of employment opportunities.

AI Gaming Roadmap

Transition users to the new token ✔ (complete)

Implement online recruitment system ✔

Implement online certification system ✔

Implement crowdsourcing platform ✔

Recruitment system — user history browser ✔

Create bot service marketplace ✔

Identification of alternative blockchain

Write competition results to blockchain

Record certification results to blockchain

Record crowdsourcing results and feedback to blockchain

Third party integration (allow another aigaming.com style platform)

Integrate with other emerging AI bot marketplaces/ecosystems

Token Information Summary

At A Glance Token Information

Token Symbol: AIGC

Token: Ethereum ERC20 Token

Token Sale: April 2018 to June 2018

Number of Tokens: 100m total supply (unmintable)

Private Sale: 50% bonus

Phase I Public sale: 1m tokens, 25% bonus

Phase II Public sale: 3m tokens, 12.5% bonus

Phase III General Sale: To hard cap of 50m tokens

Soft Cap: 2m tokens

Community Reserve: 25m tokens reserved for future incentives to build a developer community, for example, competition prizes, marketing, compensation for champions.

Company Reserve: 25m tokens for token sale bonuses, sale costs, bounties, partner compensation, future fundraising

Use of Token Sale Proceeds: Secure and build the platform and website, including:

Recruit new users
Expand educational programmes
Accelerate product roadmap
Marketing to new customers

Legal Compliance & Obligations

AI Gaming is a trading name of Impact.ai Ltd, a UK limited liability company. In the absence of UK guidance or regulation on cryptocurrency token sales, we have taken the following steps to ensure compliance:

AI Gaming has retained an SEC compliance law firm, Thompson Bukher LLP of NY, to ensure we remain compliant with US regulations. Their extensive analysis demonstrates that the AIGC token does not pass the Howey test and therefore should not be considered a security. However, out of an abundance of caution in an evolving legal area, we have made the decision to only sell to non-US citizens, or US citizens who qualify under SEC exemption Regulation D.

We have also taken note of the guidance released by the Swiss Financial Market Supervisory Authority (FINMA), which suggests that the AIGC token is classified by them as a Payment Token and a Utility Token, but not an Asset (security) Token.

We have been in consultation with the UK’s Anti Money Laundering Supervision department of the HMRC, and received guidance on our token sale which we have implemented.