**Summer is coming up. It is about time to check what you really know about the techs that “might” replace you during your vacations.**

I hear about “technology” all the time. All finance professionals around me seems to be very excited about it, and eager to take full advantage of it, or at the opposite, scared it might eventually make their job irrelevant.

But when I scrape just a little bit under the surface, for both of them, I realize that most of them often don’t really know much about it.

**I don’t know much about it either**. And this is why I constantly read and watch on various tech subjects in order keep up. This is hard and time consuming but I do think it is unavoidable, and that everyone should do it.

So here is a quick and modest ( I do not pretend to be an expert in those subjects ) little quiz that will check your “basic” knowledge about the 3 main “hot techs” of this spring in service driven industries : **crypto & blockchain, artificial intelligence and quantum computers**.

It won’t cover much, it won’t give you every information you need. But I do hope you will find it interesting, and that it will help you get interested and further pursue a better understanding of those subjects and the various myths that surround them!

Oh and don’t be scared about the first question ! Every bit of explanation is in the answers…

## Theme 1. Blockchain & Crypto

**Question 1.1. The algorithm at the very base of the bitcoin blockchain is a**** …
**

- Symmetrical encryption algorithm
- Asymmetrical encryption algorithm
- Hashing algorithm
- What the hell are you talking about ?

**Answer 3 – Hashing algorithm**

A **hashing algorithm is a function that transforms a simple text into a very complex one of a fixed length**. The main idea is that while it is impossible to find the starting text (usually a password or a file) from the ending text, the same starting text will always be transformed into the same ending one. This allows to store your passwords securely (kind of… just hashing them is not enough actually) by keeping just the “hashed” version of it. When you type your password later, the website just have to hash it again and check that it match what they have.

One other very important characteristic of a hashing algorithm is its “avalanche effect” which means that** a very small change in the input should change the ouput entirely**. If you want to try this out in reality just go here !

The bitcoin blockchain is constructed on the SHA-256 hashing algorithm ( SHA for *Secure Hashing Algorithm*). Bitcoin miners are actually trying all day to find “passwords” whose hash start with a certain number of zeros in order to “sign” new blocks.

**Question 1.2. How secure is this algorithm right now ?
**

- Quite secure, the SHA256 algorithm could be cracked in 1 year by my computer.
- Very secure, it would take 10 years.
- No problem it would take more than 1.000 years.
- This is so secure it is hard to express it.

**Answer 4 !**

The total number of possible combination of the SHA-256 algorithm is 2 power 256… which means it is NOT possible to just try all possibilities. **It is A LOT of possibilities**…

A very good explanation of this can be found in this great video, watch it.

So the algorithm is** incredibly secure… to brute force attacks.**

**BUT it might not stay secure in the future.** This is really a matter of computing power. The algorithm is secure for now, but shall the computers be faster in the future this might not stay true. Moreover, it is actually secure for long and complex inputs, and not at all for simple/short inputs… so **keep your passwords very long** as you where told already !

**Question 1.3. Which one of those is the lesser of blockchain problems ?**

- Its distributed ledger principle implies weak confidentiality of transactions.
- It is designed to use larger and larger computation power, and therefore energy.
- It can be kind of slow.
- It uses a consensus algorithm and therefore is theoretically subject to consensus attacks.

**Answer 1 – weak confidentiality of transactions
**

Actually it is quite the opposite. The strong confidentiality of the cryptocurrencies blockchains, makes it **harder for governments to trace illegal activities and especially money laundering schemes.** It is still possible to trace bitcoins and matching money outflows, but is makes it a lot harder (kind of an anonymous Swiss bank account if you want an old shool comparison).

2 is very true, and problematic, especially for bitcoin where the total electric power needed to “mine” (finding the “passwords” that have the required hash ) has gone crazy. **This is a grave ecological issue**. And the way the blockchain is designed can only make it worse in the future.

3 is true especially for large blockchains where there is miner congestion. For example a **bitcoin transaction can be processed from a few minutes to a few days**… which is very very slow compared to the current speed of financial transactions that can happen numerous time within a second.

4 is theoretically true, and a real concern. As blockchain are not “owned” by anyone but their participants, it is possible for a very large group representing more than a large part all participants to **conduct attacks by simply changing the rules of the blockchain**. While this is highly improbable for large blockchains like bitcoin, it is more problematic for other smaller chains, mainly owned by their creators.

**Question 1.4. One interesting potential application for blockchain is decentralized financial markets. Which of the following advantages would it grant ?
**

- It would reduce the cost of investing.
- It would give more control over one’s financial data.
- It would help stabilize the financial system through better equity price discovery.

- It could allow more direct financing of venture projects.

**All but 3
**

While 3 ( better price discovery) is not proven false, **there is absolutely no evidence of it**, and “price discovery is actually already pretty good in current markets.

The **main advantage of “blockchain financial markets” is that their would be no intermediaries** : no banks holding securities, no brokers, no exchange companies, no clearing houses… This means that 1 and 2 would be very true. The costs associated with investing would go down and everyone would have control over all of its financial data, banks for instance won’t be able to retain data as they do now.

4. is true too but not really implemented yet. By creating a blockchain of equity shares or bonds, **companies could raise capital differently and potentially more efficiently**. But this is not that easy and as for now even crowdlenders still haven’t really managed to use it. This, while interesting, do meet legal, financial and technical difficulties.

## Theme 2. Artificial Intelligence

**Question 2.1. Which of the following**** is not a sub-domain of artificial intelligence?**

- Neural networks
- Markov’s networks
- Fuzzy Logic
- Heuristics

**Answer 2 – Markov’s networks
**

**There is no such thing as the “Markov’s networks”**. There is a however a Markov’s chain mathematical concept that can be used in AI.

Fuzzy logic is a sub-domain of AI which aims at **handling problem where human reasoning is based on vague values and not numerical value.** The base example is the bath temperature automatization one : If the water is “cold” open “more” the hot tub, is it is “hot” close it “a little”.

Heuristics are **imperfect estimations of the progress toward a specific goal** and are used in solving very complex problems. Those of you how read my previous post about AI and video games know it.

And everyone of course knows that **neural networks** are the season’s “hot” concept for AI.

**Question 2.2. How does an artificial neural network work ?
**

- It reproduce exactly the way a human brain works.
- It splits problems into multiple sub-problems (layers) in order to solve them.
- It recognize patterns of situations and find solutions fitting them.
- We don’t really know.

**Answer 4 – We don’t really know
**

And that is the beauty and the biggest problem of it ! We don’t need to for them to work but we cannot really optimize them without.

In fact it is probable some kind of a mix between 1 2 and 3.

Neural networks are somehow designed with similarities with the human brain, but up to a point. They are designed with large numbers of interconnected “neurons”, and the connections between them are tuned with existing data in order to best fit inputs with expected outputs. **However human brain have a lot more neurons layers**…

The classical application **example is image recognition**. When you give a cat image to a neural network it will only have (like you) as inputs the colors of the various part (pixels) of the image. But with sufficient experience ( examples ) it will be able to slowly recognize patterns, or specific colors, or shapes, that have a greater probability to be found in cat images.

**The way they find the “right solution” is unclear mathematically** speaking but they do work. This is however a huge limitation to the optimization of neural networks. The lacks of theoretical understanding of it makes it very hard. For example, we don’t really know the best number of neurons and/or neuron’s layers to use. This is very empirical right now.

If you want to learn more on the subject you really should watch this amazing tutorial video about neural networks.

**Question 2.3. What is not always needed for a neural network to learn ?**

- Computing power
- Real-world data
- Human input

- Smart people

**Answer 3 – Human input
**

Neural networks do always need, in order to learn, a lot of computing power, a lot of real world data and of course very smart people to make it happen.

**What they do not always need is human input.** This is because some neural networks are actually improved through what is called “unsupervised learning”. Which means the computer learns without us… scary…

To get back to the cat image example : you can “give” the neural network a lot of pictures and tell it on which one there is a cat. It will work, and this is called “supervised” learning.

But you can also just give it the pictures with no label on it. It will allow the network to spot groups of pictures with similarities, “clusters”. It might then be able to recognize cats without knowing that they are “cats”.

**Question 2.4. Which one of the following fields is not currently subject to neural network applications ?
**

- Data-mining

- Food processing

- Law
- Medecine
- All of them

**Answer 5 – All of them…**

However the extent of its use do vary a lot between them.

1 is true. Data-mining is the process of finding meaningful information inside a enormous set of data (often correlations). Neural networks classification or “clustering” capabilities allows them to do so. Warning ! data-mining can also be a statistical error. Read this to be informed on the subject.

2 is true. The creation of digital “noses” that can recognize odors though neural networks allows food manufacturers to better control quality and detect anomalies.

3 – Is also true. Robots are now tested on automatic contracts evaluation for example. However this is still limited to very standardized tasks. Lawyers still have some time.

4 is true. Especially in fields like cancer risk prediction and diagnosis.

## Theme 3. Quantum computing

**Question 3.1. Which one(s) of the following organization is not conducting research and design on real world quantum computers ?**

- NSA
- D-Wave
- IBM
- Cyberdyne Systems

**Answer 4 ( obviously… and gain a bonus point for the movie reference ! )**

D-Wave is indeed a real company designing quantum computers. IBM and Google are also conducting research, and there are more of them. **Those “computers” are still very expensive prototypes right now but they do exist.**

**There are still a very long way from being available to the public** or even to help replace you though.

**Question 3.2. How could quantum computers theoretically surpass traditional computers ?**

- They should process basic computations a lot faster.
- They should use a lot less energy to do so.
- They should process some mathematical problems more efficiently.
- They will look better.

**Answer 3 – They should process some mathematical problems more efficiently.**

Actually answer 1 is really wrong ! Pure quantum computers would be extremely bad at solving basic computations.

But they should be extremely good at some very specific maths problems.

**What they are supposed to be good at is “optimization problems”,** a very narrow subset of mathematical problems, because their computation power will increase exponentially with their size, which is not the case with traditional computers.

That is why quantum computers will probably be “hybrid computers”, which means they will have a traditional computing part and a quantum one, trying to use the best of both.

** **

**Question 3.3. Why could quantum computers be a potential threat to internet security ?**

- Their computation power would allow them to “brute force” encryption algorithms.
- They could be able to “crack” certain security related algorithms.
- They would allow the creation of unbreakable algorithms.

- They would allow teleportation.

**Answer 2 – They could be able to “crack” certain security related algorithms**

Internet security algorithms ( the “functions” that are for example turning your password into something very very complex, like the SHA256 algorithm ) are created by maths problems that are not solvable with current computers. This means that the only way to “crack” them is to try all possibilities and see if it works. This is called “brute force”. And it does not work when dealing with sufficiently long passwords.

But, due to the way quantum computers are “supposed” to work they might be able to find the solutions to those specific problems, **thus reducing considerably the time needed to crack those codes, making them possibly useless**.

This is no surprise that the NSA decided to launch a quantum computer project…

But don’t worry too much either because this is like the weapons vs armor problem, and people are already working on quantum-proof security algorithms.

( answer 4 is actually not that stupid as there do exits a “quantum teleportation effect” ! See here )

**Question 3.4. Which of the following is not a actual problem in the practical design of quantum computers ?
**

- Maintaining quantum coherence long enough.
- Physically scaling the number of qubits (quantum bits) to a useful level.
- Creating qubits that are small enough to fit in something smaller than a truck.
- Finding a reliable way to detect qubits errors.

**Answer 3 – Creating qubits that are small enough**

Well this is kind of tricky. Current designed “qubits” are obviously very small no problem with them. But the whole computers are still very very big indeed, and might need some miniaturization. **But there are way more urgent problems to solve before that, among which are** :

1 – **Maintaining quantum coherence long enough** is indeed a real problem. Without it, no computation is possible. Materials ans designs of qubits is far from standardized, but for now maintaining coherence (the “state” of the computer in which it does have a usable quantum behavior) often imply cooling the qubits down to near absolute zero and isolating them from any source of de-coherence. Not something you will be able to do in your laptop soon…

2 – The REAL advantage of quantum computing is in its ability to exponentially scale its computing power with the number of qubits. Under a certain number of them (around 50, but this is very uncertain actually) there is no point in it, traditional computers will always be better, whatever the computation. **This number is called the “quantum supremacy point”**. Unfortunately linking numerous qubits is very far from been easy.

4 – This will also need a lot of very smart thinking. IBM among others work a lot on this specific problem. Normal bits are subject to physical errors ( oups it is a 0 and it should be a 1) but corrections are mastered. **Qubits are very subject to errors do to their fragile quantum state**. A single temperature fluctuation can incur important variations that are not planned. Spotting and correcting errors is very needed to construct a reliable quantum computer.

Conclusion for this question : don’t worry too much about quantum computers, **they are not ready yet.**

I hope you did well, and I hope that you did learn a little and got interested in those subjects, which was the very goal of this.

For those you do have a lot more knowledge than me on those topics, please correct any errors I might have made in comments. I will gladly learn and update this post.

** **

## One thought on “Spring break hot tech quiz”