AI and Blockchain for Everyone!
This post was originally published on blog.thu-tiba.org
This Wednesday, TIBA held its semester-opening activity on the Tsinghua University campus. Co-hosted with TAIS (the Artificial Intelligence Association of International Students from Tsinghua), the goal of the event was to learn the essential ideas behind Blockchain and AI. We were honored to welcome more than 120 Tsinghua students from countries all around the world.
Blockchain
The first talk was a brief introduction to Blockchain by Peter, president of TIBA.
Essentially, blockchain is a tool for trustless bookkeeping on a decentralized network. Data records can be anything from financial transactions through land ownership records to AI models.
These records (transactions) are packed into blocks, which reference each other through their unique identifiers called hash.
Combined with Proof-of-Work, Bitcoin’s blockchain creates a system where it is extremely hard for attackers to change the ledger.
We also learned that blockchain can do much more than *just* keeping track of financial transactions. Using smart contracts, we can write code to define the data we want to work on and the rules operating on this data.
This enables a number of other use cases, including crowdfunding, supply chain tracking, micro-payments for self-driving vehicles, and even decentralized autonomous organizations (DAOs).
The key takeaways are:
- Blockchain puts existing record-keeping systems on a new, decentralized foundation where participants do not need to rely on a central authority and do not need to trust each other.
- Blockchain is not Bitcoin; Bitcoin was the first system to introduce blockchain but by now we have many other systems and new use cases for the technology.
- Blockchain is not just for tech people: it can be applied in many different areas.
AI
Following the talk on Blockchain, we got the chance to learn about the main ideas behind Artificial Intelligence, presented by Fede, president of TAIS.
Fede started with a claim: ”Blockchain is simple but non-intuitive. AI is complex but intuitive.”
We learned that AI is essentially a new programming model.
Before, we used to write programs that operate on an input and produce an output. With Machine Learning (which is a sub-field of AI), we use the input and output (data) to produce a program.
This is possible as we have more data and computing power than ever.
But we also have harder problems to solve — problems where traditional programming (hard-coding rules) proves insufficient.
Machine learning is basically about learning an unknown function.
The input of the function is the features — attributes of the input data.
For instance, the speed of a car, humidity of the air, the gender of a person can all be features used in different models.
The output can be one of many things. It can be a class of objects when we talk about classification. Examples of this are spam filters where the two classes are SPAM and NOT-SPAM, or speech recognition where a word can be one of a few thousand items in the dictionary. Outputs can also be numbers when we talk about regression. For instance, we might want to predict housing prices based on historical data.
Classical machine learning (that uses approaches like decision trees or SVM) is often insufficient to deal with the complexity of our world. To solve the most challenging problems, like self-driving vehicles, we started using deep learning and deep neural networks. These can do complex things like diagnosing patients or change picture styles. But there is a trade-off: these models are much less transparent to us — their complexity makes them hard to reason about.
Not even mentioning the problem that distinctions in human-land are often not clear even to ourselves. Let’s stay you are driving and you spot a person holding up his hand. What does this mean? Is he signaling you to stop? Is he a hitchhiker? Is he about to turn?
Which all its challenges and potential, AI about to have an impact on all aspects of our lives.
Pizza
Having learned so much, we needed some time to digest it all.
And what helps more than some pizza and discussion?
(We would like to thank CoinNess for their support with pizza.)
If you want to learn more or join our team, make sure you come to our next event! On the 19th of September, we are going to welcome two Beijing-based startups that aim to combine AI and Blockchain in various ways. Looking forward to seeing you all there!
To learn more about TIBA and join the discussion, scan the QR code below!
This article was written and edited by Péter Garamvölgyi and Federico Zaiter. Photos taken by Dylan.
