Category: AI & Automation

The human mind is fascinating in so many ways.

I often like to invision the brain as an input/output machine with some internal loops to self-reflect and generate its own internal inputs. Inputs to the brain come from our senses – what we see, hear, smell, touch. Those inputs will physically modify your brain as new memories are formed and core-beliefs are established. As a result of those inputs your brain will produce an output, you’ll move your body, say something, or think something.

Every detail of those outputs from big ideas to micro variations are influenced by the inputs to your brain in the first place.

Most people, including myself, don’t take conscious and critical consideration about what inputs we provide for our brains to process and build memories from. We passively consume TV, read the news and listen to podcasts or radio for the emotional and entertainment value.

What if we specifically fed ourselves the kinds of information that would align with our goals, or placed ourselves in situations to learn and practice the skills we need to get better at the things we deliberately want to pursue. This kind of intentionality would be a game changer if applied consistently for months or years.

So, I’m encouraging everyone to take care before they train their brain with random inputs from news sources, social media, video games, music, conversations, or TV shows. If training your brain on something is not going to be of benefit to you, decide to do something that is instead.

Last night I was attempting to learn a little bit of Chinese after being inspired from a series of documentaries and travel vlogs (ChopStick Travel) that I’ve been taking in over the last 3-4 weeks.

Anyway, I opened up DuoLingo and was quickly disappointed. Then I tried Memrise, which was slightly better, but it suffered from the same issues.

Most critically, these apps fail because they have no variability. Every time you hear the recording of “nĭ hăo” it is the exact same one. Every time you see the script written “你好” it uses the same font.

Deep learning is a mathematical approximation of how the neuron’s in our brains work. At least when you compare to very simple brains like those in worms, current AI tech is a reasonably good simulation. I think it’s reasonable to assume that we can gain insight into the human brain from things we learn developing AIs.

When we build these simulated neural networks how they are trained is critically important to how they perform. For instance: training a computer to understand spoken language requires an immense number of audio samples (Mozilla’s open data set is currently at 12GB of compressed audio). When training an AI to do visual character recognition it takes thousands of samples to get a good working model. Elon Musk has said they need 1 billion miles of recorded driving to get a reasonable self driving model. AIs need a LOT of data to train on.

Granted the human brain is vastly better at learning than current Deep Learning AI algorithms, but we are not so much better that we can learn new words with a single sample. Hearing a word said in one way, recorded on a sound stage will do very little to help recognition when you hear it in person outside at a loud food market, or with a different accent, even a cadence change might throw off your comprehension.

If you came to an AI expert to get a voice recognition system trained and you had spent $50,000 at a sound studio getting one perfect example of each word to use as your training data. You’d be laughed at for wasting so much money on something so useless. It is preposterous to train AI on single sample sizes, but that seems to be what we expect to happen with people.

I wish there was a language learning platform that took the approach of compiling a minimum of 100 samples of each word (at least at the beginning levels) ideally taken in-context with video so you can see the facial movements and pick up on body language. It could initially seed the library of data with clips from YouTube or movies. Over time it could ask users to contribute snapchat style short clips of them saying words/phrases in their native language to help others on the platform. A particularly memorable clip might be the one that cements the word in your memory, and hearing all the variations will help train your ear to recognise the word. It could use image recognition AI training sets avoid learning a mental translation process that slows down fluency.

Using the same kind of data we use to train a Deep Learning algorithm but use it to train humans would be super interesting. It might unlock insights in psychology, education and computer science.

As an endeavour it would be awesome to have 10 new people try the app every month and do a 1 hour session on-site with a new language. Then see how much they get through a conversation with a native speaker afterwards. Optimise the app over time for real improvement with comprehension and perhaps conversation. Language learning is an ideal case for testing the limits of how quickly humans can learn something new.

Perhaps a billion dollar idea if only I had the capacity to pursue it. (the amount of user contributed data could be a gold-mine if it worked)

Over the last week I ran into a difficult problem to solve with my B

lockagram application that I have been working on for the last few months. It required some deep thinking.

I spent three days with some sensory deprivation to really focus my thoughts and avoid even the slightest distractions. An eye mask and hearing protection while lying down on the floor of my office. Strange but effective.

After all this, I think I have had a real breakthrough for a way to create an innovative new proof of work algorithm geared towards a block chain for identity. Unlike Proof of Work used by Bitcoin this will require no wasted computer cycles and so it will be very power efficient.

Yesterday I started to work on a white paper to describe how this new block chain application will work and have started to develop out the proof of concept with code. Hopefully there are not too many roadblocks to develop it.

Amazing that this kind of thing just wouldn’t be doable in a job setting yet can be super productive. Finding solutions to really difficult problems can require a lot of thinking, and deep thinking is hard to do on a timeline.

This whole thing is a bit of a tangent from my main application. Trying to find a way to avoid having spoofed accounts, and fraudulent users led me down this rabbit hole. It’s a luxury that I can take the time to see how far the rabbit hole goes

It’s been a while since I last did any homebrew beer but for some reason I had the motivation to put on a new brew when I got back from vacation.  Nothing fancy, just a simple kit.

The one new piece of equipment I got was a conical fermenter.  The Fermentasaurus is a pretty neat bit of kit and makes things much easier.

With this I can avoid a full re-rack step. It’ll cut the amount of cleaning and number of steps in half compared to the way I was doing it before.

If things turn out OK, I might try to build another little project to help me avoid needing to move things at all. Ideally I’d like to get a good fermentation chamber built in the garage and perfect a lager recipe. Minimizing as much of the hassle as possible.

After understanding the basics of handling money and what is a broker and what do they mainly do, I have now been getting serious about crypto-currencies and blockchain technology at an accelerating pace for several years now. The more I think about the implications of a decentralized protocol for money and the exchange of value the more I see it as a disruptive technology that will shake the foundations of how the economy works.  It is more of a shift than you can imagine.

The first thing to note about this technology is that it is un-stoppable.  Now that it exists, there is simply no way to put it back in the box.  There is no feasible way to block the traffic on these protocols over the internet and there is no way that governments or banks will be able to permanently prevent or lock out the ability to sell crypto-currencies back into fiat.  There is no future where bitcoin won’t exist – even if in 50 years people don’t use bitcoins it will be kept alive for the nostalgia.

Secondly, while bitcoin is almost 10 years old now, it has only recently become more than a toy for dedicated hackers.  Realistically it’ll take several more years of development before normal people use it.  This is a long term play for the future and we’ll continue to see improvements happening going forward.

When we extrapolate out what we expect to see in terms of properties of future blockchains things get crazy.  Transaction costs could eventually approach electricity costs, which should be close to zero as more efficient consensus and validation algorithms are developed. At the same time expect major improvements to the transaction throughput as various strategies are used to scale to millions of transactions per second.  With low cost and high scale it becomes possible to create entirely new paradigms about how money is spent.

Imagine I am an author with a collection of books published through a publisher and available for sale in various e-commerce and retails stores.  Each book has a complex set of royalty agreements that determine percentages for the editor, co-authors, artists etc that had a hand in writing the books.  In the current state of the world I would probably get a quarterly cheque.  but the blockchain future allows this to be a real-time stream of money flowing into my accounts. When someone buys the book, the retailer takes their percentage and pays the wholesale price into the publishers smart contract which calculates the royalties and the appropriate amounts cascade into beneficiary accounts. Thousands of retailers funnelling money into a handful of smart contracts and finally to me.  End to end completely automated and happening in just a few seconds – without banks, or credit card companies being involved.

With peer to peer micro-transfers it becomes possible to get rid of bookending some types of transactions. When you fill up your car with gas you start by swiping a credit card to pre-authorize a maximum amount to spend, and after filling the actual amount is charged to the card. If instead you created a transaction for every $0.01 worth of gas then it becomes technically possible for the car to pay the pump directly.  If the car had a computer that could talk to the pump and that computer had the authorization to purchase gas then the driver wouldn’t have to deal with payment directly at all.  You may think “big deal”, but if the sharing economy takes over with cars, then avoiding the complexity of how to pay for the gas/electricity is an important thing to solve.

If you think beyond simply money then things can get more interesting.  digital ownership can be tracked and transferred. Unique digital goods can be created which can’t be copied.  Identities can be verified without even seeing an identity. There are interesting things that can be done with cryptography when building on top of these distributed systems.

This stuff is going to throw the finance sector into a bit of chaos for the next decade or more.  What do you do with calculating capital gains if you are atomic swapping between a dozen different crypto tokens and currencies with thousands of micro-transactions per day?  Will law enforcement lose the ability to seize the money of drug cartels, or countries that don’t pay their bond holders?  Will privacy coins make the economy as opaque as a GPG encrypted hard drive? Stay tuned.

Economists are going to have to re-think some significant assumptions in a new world  where trust is not required for efficient co-operation and where machines can become their own independent actors within the economy.

There are just so many things to be excited about in the future. It’ll be fascinating to see how it all evolves and try to be a part of as much of the coolest bits as I can.