Blockchain Based Control and Safety of Artificial Intelligence

Buzzwordy title alert.

Although there were many individuals worried about recursive self-improving AI, the alarms weren’t really sounded until Nick Bostrom wrote Superintelligence. For those readers who are unfamiliar with why superintelligent AIs, AGIs for short, might be scary, they can look at my notes or this post here. Long story short, an AI that is vastly more intelligent than us that isn’t aligned with our interests may decide something that isn’t in our best interest. 

The oft-quoted example of AGI, aka superintelligent AI, gone awry is the paperclip maximizer. While this example doesn’t exactly capture all the nuance, one can get the gist of the problem. An AGI is created whose sole goal is to create as many paperclips as possible, since it’s so good at its job, ends up killing all humans and turning all matter into paperclips. A more “human” example of an AGI gone away is a corporation, aka Enron or any oil company. Cash flows and profit, the internal metric of success or objective functions, they use becomes divorced from their original purpose of creating a good for society. Bitcoin and other cryptocurrency networks also represent some kind of recursively improving organism with no clear point of disconnect and have some individuals worrying about blockchains and AI. AGIs gone away would represent the principal-agent problem on steroids. You could well argue that Bitcoin, or cryptocurrencies are a version of this paperclip maximizer, especially the Proof-of-Work variants. 

The basic assumption that researchers in the field make is that AGI is going to happen someday. If not 15 years away, less than 100. 100 years in the course of the universe is nothing. Therefore, solving this problem of defining an objective function, or guardrails for an AGI is of the utmost importance. Sadly, this isn’t quite incentivized today. However, the work that has been done can be summed up as such:

  • Alignment: Making sure its objective function doesn’t kill us. Work that I’m most familiar with is coherent extrapolated volition and approval directed agents.
  • Capability restraint  For example, an AI that is air-gapped from the internet can give just yes or no answers, aka becoming a genie.

However, Bostrom presents another idea on AI control that I think doesn’t get enough coverage. In a few short words "make the objective function tied to the acquisition of some cryptographic token". While this seems unintuitive at first, it becomes akin to us trying to earn money, or dogs doing tricks for doggie treats. In the original proposal, Bostrom proposes to use a centralized cryptographic token managed by scientists. Superintelligence was published before this current hype cycle as well as theoretical work on new cryptographic primitives had begun. During that time, there’s been a little bit of fervor over how blockchains can positively increase the capability of artificially intelligent systems such as Computable by providing more data sets, not much has been written about the safety side.  (No surprise there). Here are some specific high-level proposals that can be stacked on top of each other to control and align agents.
  1. Use a decentralized cryptocurrency as reward function. This one is straightforward enough. Using centralized cryptographic tokens as the goal suffers from the same reason that centralized cryptocurrencies didn’t take off. They introduce the same single point of failure. If a scientist is somehow held at gunpoint by an AGI, he or she will probably hand over some tokens. It’s much harder to hold a network of miners and anonymous token holders at gunpoint.
  2. Instantiate an AGI as a DAO. This allows this entity to operate trustlessly, which is a double-edged sword. This allows the AGI to sustain itself and operate with or without supervision. But it also keeps an auditable trail of where and when the objective function cryptocurrency was added to the specific address. 
  3. Define reward function as a smart contract to be executed trustlessly. This is where it starts to get a little harder to conceptualize. We can state in plain English what something is. This matters for reinforcement learning agents. Objective functions in terms of Starcraft or Go, are simply to win the game. However, we may want to iterate/check up on the operation of the AI, and update an objective function as we go on, and not let the individual agent be able to change any part of the objective function. Then, use a widely distributed governance token, so pseudonymous actors can allow for changes to this governance token. Keep identities private so that the agent isn’t able to harass/bribe them. Monitor past voting behavior, by adopting a trail of “reputation” for voters to check for any bribery, this can also be determined on-chain.
  4. Use curve bonded tokens to get rid of “take over attacks”.Curve bonded tokens have programmatically defined prices for minting and redeeming (and then burning) a set of tokens. To perform any goal, the agent is probably going to have a lot of cash on hand. What if he tries to buy up a supply of governance token? That would be bad, then it could change its own objective function. To prevent this, we can set the curve for purchasing tokens at an absurdly high price as more tokens are minted. Corresponding, we can set an extremely small sell-out price to disincentivize any sales.
  5. Use TCRs (or some other game theoretically sound) ranked lists to tokenize “human values” and direct an AGI to optimize for that set of “human values".The previous example talked about defining a goal in terms of ETH held. That would be easily calculable if the goal of the agent was to maximize the NAV of its investment portfolio. However, as we know today, defining something just in terms of money can lead to some perverse outcomes. If the means of money become the ends, then that leads to greedy short-term actions that can be taken by agents.
  • Instead, we might want to optimize for human well-being. How do we define this on chain, so this measure can’t be hacked by an autonomous agent? We utilize decentralized stake-based rating games, namely TCRs with a curve bonded token for staking. You can read a little more about TCRs here.
  • Back to representing human well-being “on-chain”. First, we have to define how this is defined in the real world. Various NGOs and ratings orgs track things like the HDI, Human Happiness Index, and GDP per capita. These are top line objectives that countries may try to aim for, through actions that make individual citizens happy. Of course, countries are free to ignore these ratings as well. However, autonomous agents won’t be if their objective functions are locked down.
  • So how does that tie into the blockchain? These indexes have a large self reported component right now, and TCRs are good for encoding intangible and subjective information into hard economic terms. By creating this list that might be composed of “happiness”, “wealth for humanity”, and “sugar, spice, and everything nice”, we might have the agent take off-chain actions that benefit humanity.

The largest points of failures would seem to be the voters, especially if they have their identity revealed. Perhaps we can have less intelligent agents that vote on issues for the most intelligent agent, each with their own objective functions that need to be modified. With any organization or incentive structure, there always needs to be a balance between being able to change something and not letting the wrong actors change things. I think this game is especially fun to play when thinking through an actor that is vastly more intelligent than I. 

Early Adopters of Crypto

Attention is the most scarce thing in the world. On a macro level, the world is awash in capital. Interest rates in countries are below zero. However, within our daily lives, there are always thousands of things competing for our attention. A question I like to think through is, where are the early adopters focusing their limited attention. Chris Dixon says it’s people messing around in garages building something. A revised question along those same lines is: 

Which nation/market is an early adopter of technology? How do their market dynamics predict what might happen in another geography?

First, a little theory. The world is a connected graph of people. Word of mouth is the thing that really gets people to adopt products. Facebook decreased the six degrees of separation down to around 4.5. However, among this distribution of connections between people and connections isn’t even. When we think of information flow, it’s more of a uni-directional graph. This means that person A can influence person B, but not usually not vice versa.

When we think of how information spreads, I think of a tinder over a dry terrain. While something doesn't spark 100% of the time, but when it does, there's the potential for a cascade of "catching fire". Within a network, there are early adopters and late adopters. These people are differentiated by personality traits, sources of information, and levels of connectedness in both the upstream and downstream direction in terms of where they get their information. I usually split the adoption curve into three sets of people:

  • 1) people who do things because it is novel or cool. This is an intrinsic motivator. These are early adopters.
  • 2) people who do things because there’s an economic need. This is an extrinsic motivator. These are middle adopters.
  • 3) people who do things because everyone else is doing something. These are late adopters.

So now that we have that out of the way, this is my current mental model for crypto adoption.

I am increasingly looking towards Asia for technology and more specifically Korea for cryptocurrencies. Due to special features in what their graph looks like, they have interesting winner take all dynamics as well as being early adopters. Information spreads quickly because of the connectedness and centrality of its social graph. The whole nation using Kaokao, has high-speed internet access, a high appetite for novelty and coolness, very tight-knit business communities, and have historically been early adopters of new technologies. Before the States got around to these things in Web 1.0, Korea was already on top of camera phones in the early 2000s, playing MMORPGS and other things, and over the top streaming (aka Netflix).

Bill Gurley and associates caught onto this trend and planned a trip to Korea to see what might be gleaned from this market. What resulted was a sharpening of their thesis around Social, Local, Mobile. When the iPhone hit everyone’s hand in 2008, we had the confluence of the internet, GPS, and camera in every pocket. And the rest is history, that Benchmark fund invest in a plethora of internet hits most notably Uber and Snapchat.

The current environment for Korea is pretty telling. 30% of South Korea owns or holds some sort of crypto, past the tipping point for widespread social adoption. When the regulators tried to shut exchanges down, HODLs raised their voices. I’m excited to see how individuals interact with token powered protocols as usability and scalability allow us to fall down the Marginal Benefit Curve of cryptocurrencies. While we’re still stuck at the store-of-value and the speculative era of cryptocurrencies, that should change soon.

Even now, as staking protocols begin to proliferate, crypto holders are looking to gain an edge in earning incremental token. We should start to see use Vest and Compound.Finance gain adoption as the usability of protocols begins to drop.

I’m personally not as bullish for developing countries as leading indicator as early adopters. As weird as it sounds, they need cryptocurrencies too much. My mental model for early adopters are the ones that like toys, the weirdos, the rich people and more that are willing to accept the flaws in the product. There’s something about intrinsic motivation as opposed to extrinsic motivation that drives the sickness and retention of a product/technology. I would much rather look towards the high-risk tolerance ICO investors than look towards traditional business and crypto “enterprise alliances”.

Merging Chains

Written by Niraj and Dillon

If you posit that bitcoin has a network effect (the more people that use the currency to transact), the more valuable the coin becomes. The more valuable the coin becomes, the more users you get and the higher the network effects. Additionally, if longer chain history means better security and more miners mean better security, in the long run, is there a way to increase the network effect by merging chains?

Right now, we've only got people doing forks. Forks are important, they allow for experimentation on the rule sets, however, they potentially may reduce the overall network effect of any single token. Forks are really good because they align incentives between the people who have already done work on the master chain. In the example of ETH and ETC, we argue that it's a feature that the Ethereum Foundation automatically held ETH and ETC without their permission. They could gain in economic value of another development team. The new development team wins because they get an input customer set, the set of pub-priv key pairs that already holds ETH. This is a subtle shift in incentives, we'll write more about this later...

While we're not advocating for a maximalist approach, the idea that there should only be one token ever, it just seems like there needs to be a process to merge chains just as there is the process of forking. There is an argument to be made that "Core" or the Foundations bearing the base token name centralize development resources. In BTC and ETH respectively, only 5 and 2 developers make up the majority of commits. Forking seems to have to have become a way for talented devs to work on protocols. Just look at LTC and @satoshilite.

Additionally, we see that experimentation has been a net positive for society in other areas. Allowing for experimentation and merging isn't limited to blockchains. Just to name a few:
  • Policy experimentation within a federal system of government. I.e. adoption of a precursor to the Affordable Care Act before it became national law.
  • Startups as new entrants that can be acquired or grow to be large companies.
  • Spin offs from large corporations. Standard Oil became several smaller companies and Rockefeller was richer for it.
  • Mitochondria being swallowed to become the powerhouse of the cell.

In blockchain terms, you could conceive of merge mining as extended uncle resolution. In the GHOST Protocol, the individual uncle hash power is added to the winning block's score. Uncle miner still is incentivized, they get some proportion of the block reward. Likewise, people who contribute to the "losing token" are still incentivized. When you think of merging chains, you're still incentivizing a smaller chain's absorption into the larger chain. While protocols can directly implement the necessary hard/soft forks to include the rule set change of a fork, they won't have the now differentiated userbase etc.

How to Do Merges

There are two methods for potentially doing a merge for tokens (and probably more that we haven't thought of).

The first method is pegging a token A to token B.
  1. Agree on a price/exchange rate for A:B
    1. Oracle to determine price
    2. Hash power signaling/ratio
    3. Market pricing on exchanges
  2. Hard fork both protocols to have the same block + rule set
    1. Enforce a specific block height for the rule change, include the pegged price ratio
    2. Price converges
  3. Before the rule set is implemented, people are free to trade out of token B
  4. Allow for atomic cross chain swaps
    1. Using Decred or 0x → hard code this into the rule set change
The second method involves one chain "absorbing" the value of the other. Meaning that token A remains and token B is never used again.
  1. Agree on a price/exchange ratio for A:B
    1. Oracle to determine price
    2. Hash power signaling/ratio
    3. Market pricing on exchanges
  2. Acquire buy out funds for A to purchase B
  3. Post a public address where all B tokens can be sent to
  4. Before the rule set is implemented, people are free to trade out of token B
  5. Burn the B tokens, each B token holder, will get the amount of agreed upon amount of token A proportion to how much they sent to the specified address

Roadblocks to putting this in practice.

Both of these scenarios involve a lot of coordination. Imaging trying to do a protocol merge without some kind of explicit voting mechanism other than hash power signaling induces a headache right away. The future of decentralized governance will definitely play a large part in how these things happen.

Also, as we see in centralized mergers and acquisitions, the larger company often has to purchase the shares of the smaller company at a price premium. We'll have to establish a better pricing mechanism beyond hash power and other matters. Ari Paul and Chris Buniske have been doing a lot of great work in fundamental valuations for this.
Additionally, atomic cross chain swaps are not the only potential way to transfer a token from one chain to another, using a protocol such as Polkadot or Cosmos we might allow for this sort of thing as well.

Real World Procotols That Could Benefit.

These wouldn't just have to be currency tokens, you could potentially also merge utility tokens as well. For example, looking at Sia and Filecoin. If Filecoin were to establish a dominant market cap and share position, it might behoove them to purchase the Sia network. An additional step would need to be taken. Individuals would need to, before they can acquire any of token A, transfer their files over to the new blockchain. Once this is performed, they can claim their Filecoin token.
  • Small cap token mergers
  • Prediction markets (Augur and Gnosis)
  • File storage markets (Filecoin, Sia, and Storj)
  • BTC variant mergers (BTC, LTC, BCC)

----

Look here as well

The Great Firewall of Facebook

Media when concentrated in a few individuals or the state has always been subject to censorship/influence whether by direct action or inaction. Western Union, China, and Napoleon are a few prominent examples. Now we have Facebook. The press lauded Facebook and Twitter when it influenced the Arab Spring, but is chafing at social media’s power now that it’s come to influence our politics in the States. 

Facebook and any social network has de-facto censorship through their use tweaking of newsfeed algorithms. Facebook is just the largest and easiest target. Of course unlike China, Facebook’s aim is not to achieve certain any certain political goal. The aim as a public company is to create long-term shareholder and user value. Therefore, any tailoring of the newsfeed algorithm will be made towards those ends. It is a bit scary that Zuck controls the majority of Facebook’s voting stock, control of the board, and further downstream of the algorithms that control our news feeds, and sheer scale of Facebook’s users. And when shareholder value seemingly comes into conflict with our user value in the most valuable walled garden in the world has a frightening influence when a country is just a medium sized audience in terms of their scale. Facebook has always been a walled garden, killing off any products (Facebook Platform, access to media, etc) that captured too much value from the all important newsfeed. Much like Facebook, the internet in China is also a walled garden. With the Chinese internet, we know the aims of censorships are to achieve political goals. 

As it’s started influence our domestic politics, there have been proposals in the tech community, within Facebook, and in the broader community to tweak the Facebook algorithm in various ways. These have been proposed solutions that I don't feel like are permanent fixes:
  1. Users proactively change feed, because people don’t shift from their default option.
  2. Tweaking the algorithm. We’ve seen with SEO and Google, this is just an arms race.
  3. We can use traditional anti-trust regulation. 
    • Iron rules of information economies, everything tends towards monopoly because of network effects and zero marginal cost of distribution
    • Smaller groups of people might become more of an echo chamber
  4. We could turn Facebook (and Twitter) into public utilities/non-profit
    • This returns to issue of who controls it. If it the government, this would always be at risk of turning into a propaganda machine.
    • If it another rich billionaire, it runs the same issues as traditional media organizations (as well as Facebook).
The core problem with Facebook, Twitter, Linkedin, is that they need to aggregate their data and users. That’s why they’ve closed off their APIs to developers. They know their graph is what makes them special. Twitter used to be open with their data, but closed it after they saw different clients like Tweetbot/Tweetdeck and like the like potential a threat to their data moat. They can’t relieve the tyranny of the algorithm

Yes, IMAP and access to emails is one example where multiple parties have access to data, and where companies can still make money.

However, the blockchain and app coins might provide a better solution to ending censorship by algorithm and still incentivizing people to create open products.. Just as we have in email, we can use multiple clients to look at our email, and as well as incentivize creators of these protocols.



 Facebook Email Blockchain Social Apps
Front End One (Money made up here with ads) One (Money made up here with ads) Many
Algos One Many Many
Token/access to data None None Money made down here with increasing data/users
Data/Blockchain Have to guard this Free to share Share freely/forkable

We can tie together a token that directly read/write access to data. The token should rise in value just as Bitcoin has risen in value as more transactions, more data is added to the ledger. When we tie the business model of the token directly to data, we don't have the same problem of not allowing Facebook to share their data. 

Just as we can view our email with multiple clients, we'd be able to view our friendship graph, the stories and links they post with multiple front ends. On the front-end, it doesn’t matter which UI/UX experience the user sees, or which algorithm the user sees*. Different entities can A/B algorithms for sorting newsfeeds. We can have ones that allow users to see fake news, ones that allow users to be exposed to more long-form content, or even ones that promote argument. Cryptocurrency based social networks can end the de facto censorship that Facebook holds over what news a user sees. And some people have already built prototypes of these social networks: Squeek.io and Eth-Tweet. To me, this seems like a potential solution that aligns everyone’s incentives. It’s a way for technology to solve problems created by technology.

Facebook represents a centralized model of social networks. They’ll still remain very important. While a great utility, it also runs counter to the spirit of the open web. Perhaps blockchain social networks can return us to the open-source past of the web, while still allowing creators to satisfy their self interest.

* Gating access to the underlying data doesn’t have to mean that the average user will have to pay access to use the service as different token distribution mechanisms can be used so that top users (which will be advertisers or celebrities) will subsidize access for the average user.  

Free Internet and Electricity (And Crypto) Everywhere

Before finals last year, I traveled to Belize to escape school. I felt the full force of the 100% humidity and the sun beating down on our backs at a scalding 97 degrees. Trouble began to brew as our car rental fell through. It wasn't turning out to be the relaxing getaway we thought it'd be. Luckily, we got a car from Pauncho's, a local car rental service, at double the normal insurance premium. We soon pulled away the airport, and set our sites on a long drive. 

Belize is undeniably beautiful. Glancing up from the road, I caught glimpses of lush greenery and huge mountains in the distance. And later in the trip, we spent time in a rainforest tree house, surrounded by the all the coos and croaks from all sides. However, this beauty was juxtaposed by the conditions of the towns we visited. I saw weather-worn houses and one-room schools deprived of access to internet. On the trip, we paid a huge premium for this privilege: $70 for a hotspot and 2GB of data. This was a luxury that many of the people I was surrounded by wouldn't be able to acquire. While meditating on that, I caught up with the connected world. 

I read about how solar energy was spreading around the developing world due to low-cost Chinese panels and about the new release of the 21 Bitcoin Computer. The "21" press release had a quote that stuck with me--"a miner in every chip and device". Sometime while reading this article, a flash of inspiration hit. I envisioned an integrated system to give access to the internet and electricity for free--a solar panel, embedded cryptocurrency miner, battery, and Wifi/3G access point. We would give the device and internet services away for free and earn money by mining cryptocurrency with free solar-generated electricity.

While we have 5 billion phones on the planet, developing nations around the world not only pay the highest costs per capita for smartphone usage but also for merely powering those phones. We know that the smartphone is everyone's gateway to the internet. However, the internet that you and I use at home is not what those in the developing world use. Phones are often unable to update their firmware because the cost of that download alone would eat up an entire month of data. Data plans can cost as much as 37% of a worker's salary per month in the developing world, and in rural areas, this is even more stark. These areas often don't have access to cellular service at all. I know this not only from months living in my ancestral farm town in China but also from this recent experience in Belize.

I recently ran a back of the envelope model to test the feasibility of this design. Thanks to increasing solar panel efficiency, decreasing hardware costs, cheap computing power, new 4G/LTE/Wifi satellites, and Bitcoin, the numbers seem to work. We could potentially give everyone in the world access to today's essential utilities--free internet, electricity, and access to a global financial system. Who knows if this idea will end up working, but the potential seems pretty great :) If anyone has any info to invalidate this idea, please do so; in the meantime, I'll be learning more about the crypto price dynamics, satellite internet, and reliability of hotspots. Then moving on to building a prototype!