Tackling humanity’s biggest existential threat. Can we tame the AI and make it even more intelligent and responsible before it reaches singularity?

Years 2015 and 2016 were landmark years for AI. 2015 saw Skype automatically translating its voice and video calls from one language to another, Facebook got it’s AI to describe images to blind people and Google got more than 2,700 AI projects going. 2016 began with big “AI Apocalypse” controversy, which continues to brew.

Leading tech luminaries from Elon Musk, Bill Gates to Stephen Hawking joined 8,000 other AI scientists to characterize AI as humanity’s biggest existential threat. Their adversaries came out calling them Neo-Luddites for touting an AI apocalypse.

The AI Does Not Hate You

Two events last month rekindled the AI controversy. The first was a book by Tom Chivers: “The AI Does Not Hate You.” The second was a grant of $188 million given to Oxford University by a US billionaire, Stephen A. Schwarzman, to study the ethical implications of Artificial Intelligence. In the announcement, he warned that technology left unaffected would “trample over certain aspects of human behaviour and human opportunities.”

“Ultimately, AI ethics and workforce management issues are going to be determined by what’s allowed to happen,” says Schwarzman.

Whether the AI apocalypse threat is real or not, the need to make it immune to rogue actions is real and immediate.

Too much power at the disposal of humanity’s not so rare rogues will never be without the apocalyptic risk. In this age of big data, Internet, AI, that’s true whether it is global warming, nuclear war, or superhuman AI.

Humans are inherently contentious and egotistical. Defending our beliefs and actions at all costs is in our genes, regardless they are moral or immoral, humane or inhumane. We do the right things only if they are convenient and non-penalizing. If doing the right thing inflicts pain, that’s not right anymore. We will defend our actions because what we believe is always right, and adversary is always wrong. That human ego is at the heart of our innate lust for power. It is also at the heart of the progress that humanity has made, and it will be at the core of destruction that could become inevitable if technological advances aren’t moderated.

Our Vision On Convergence Of AI and Blockchain

Here’s our vision of how AI and blockchain complement each other. On one hand blockchain can rid AI of apocalyptic fears and make it safe, efficient and beneficial to humanity, and on the other AI improves blockchain by taking human intervention out of its consensus algorithm and currency creation and stabilizing mechanism.

Blockchain Advances & Tames AI

Currently all AI in a sense is machine-based and not consensus based, thereby concentrating its decision making locally within the machine. AI restricted to a single isolated machine is neither beneficial nor a threat to humanity. Its benefits and threats come from its ability to connect and influence peers in the network. Connectivity benefits AI by boosting its intelligence from collective wisdom of many, but also allows bad actors to take control and harm the network at large.

We believe democratizing, decentralizing, autonomizing and anonymizing the higher level of decision making by means of consensus of the peers rather than leaving it to egotistic control of individuals will curb the perils of AI.

We do realize that there are some really serious issues with blockchain technology. The most relevant ones are speed, cost and of course vulnerability to centralization. But we also realize that much of those criticisms come from ethereum’s performance in real life use cases. We have to realize that ethereum is only the very first proof of concept of the smart contract technology. That is not to say that the future improvisations of the technology will not overcome the shortcomings.

We are forking an ultra low latency blockchain protocol that may be capable of matching 5G speeds of the future for real time communication between IoT devices, peers, and building collective artificial intelligence (CAI) algorithms for their decentralized and democratized intelligent decision-making. We believe this new networking paradigm, if scalable to practicality, may have a profound impact in almost every walk of life. We are implementing and testing some of these paradigms in a few very early use cases that are mature enough for near term commercialization.

In essence decentralization of AI not only benefits from collective wisdom of many and helps us achieve singularity earlier than projected, but the decentralization of decision making also tames the AI and prevents it from taking rogue decisions that threaten humanity in general.

AI Reciprocates

Leaving aside the possibility of existential threat for a moment, that natural lust for power is quite evident today in the way blockchain decentralization is handled today. Whether it is “proof-of-work” (PoW) or delegated proof of stake (DPOS), the current decentralizing consensus mechanisms are plagued with the menace of human lust for power.

Although blockchain, in theory, decentralizes power, it cannot completely stop human greed and craving for power from gaming the system. Pooling, syndication, cartelization are the names of the game.

Whether it is PoW consensus protocol of bitcoin / ethereum, or DPOS protocol of bitshares / steem / EOS, decentralization is far from perfect in every consensus protocol out there. Here is an extreme view of a celebrity economist on decentralized crypto:

According to Dr Roubini the estimated Gini index for bitcoin and ethereum at 0.88 is even higher than North Korea’s 0.86. In other words, crypto is less equitably spread than wealth in arguably the least equitable and most dictatorial country in the world. In comparison US is around 0.5 and Europe is at 0.35.

Conclusion

As much as blockchain can curb AI’s existential risk, AI can also help take blockchain to the next generation. Our work on using AI attempts to perfect the consensus algorithm to remove current blockchains’ vulnerability to centralization via pooling, syndication or cartelization.

AI can also potentially remove human intervention in the process of currency, creation, supply and stabilization.

Stay tuned for more.

The article is simultaneously published on Medium