College Questions: AI and Jobs

Posted by | Random Thoughts | No Comments

If AI takes over and we lose “jobs”, where will people find their meaning, creativity, etc.? Could AI become far enough advanced that we would slide down the food chain? Can AI learn common sense reasoning?

An AI question this week! This will be a bit of a philosophical, rambling post since I’m writing this at 30,000 feet on my way to St. Lucia, so no internet! This means that instead of the normal thought gather and internet searching I do for the 1st hour, I’ll be just be dumping my thoughts.

To start with, there’s a handful of complex questions here that I’ll be breaking out individually. The first is the age-old theory around machines taking over jobs and replacing humans. While I’m a big believer that even strong historical trends are not a good indicator of what will happen in the future, I do believe that, so far, every time that we’ve created a “job replacing machine”, humans end up still finding new work.

If my memory is serving me correctly, there were many times in the past where an invention that resembles replacing a human, humans have typically reacted poorly by doing things like burning the machines, lobbying to stall progress, etc. 

In the long run, it is my belief that jobs end up getting offloaded to machines. I say offloaded specifically because it implies that we task the machines to do the mundane and boring. As we enter the world of ML and AI however, machines are becoming quite amazing at doing complex tasks and exceeding human capabilities. This last point brings up where my thoughts stray from the historical trends.

I’ve seen ML/AI that can craft beautiful music, create incredible novel pieces of art, drive cars, detect cancer at a higher precision than humans, and more. This is definitely worrisome because historically highly skilled professions, such as radiology, had a significant moat around them. The jobs we historically automated away were lower tradesmen jobs. Now, we’re going after the white-collar.

In this new paradigm, what I believe may happen is that humans will exceedingly start to create tools that they can leverage in isolation or concatenation in order to compete strategically. What I don’t think machines will be able to do effectively is a win in the arena of business strategy or where human emotions need to be evoked. There’s a certain level of complexity that is hard for a machine to evoke.

Let’s take consumer branding as an example. What is difficult about really amazing branding is crafting a story that taps into the storyline that potential customers have experienced themselves. Unless the machine has a deep understanding of its consumers from a humanistic perspective, it’s going to be very hard, if not impossible, for it to craft branding at that level. If we were to dive into a specific example, we could look to New Belgium Brewing. Their craft beers attract a type of eco & socially conscious type of crowd based on their designs, the words they use, the events they hold (eg. Tour de Fat), and so much more. It’s been an embedded story in their ethos since day 1. How could a machine be able to replicate that?

Getting back to the 1st core question, I think as jobs are taken away from humans, humans will end up circling back to the arts, macro business strategy, creative fields, or highly complex macro fields. On the last point, this would look something like creating a macro-level vision for humans, such as becoming multi-planetary. A mission like that has most of its purpose rooted in the “why not do it” and the means of getting there is where humans & machines work together to get there.

On the second question of whether humans can slide down the food chain, I think the short answer is yes, this could happen but feels unlikely. It’s a yes because theoretically it would be possible but would require a ton of dependencies. Machines becoming sentient will likely take many decades and, even then, would have limited scope on what they could accomplish. The example Elon Musk usually provides is the email spam problem. If you give a strong enough AI the task to remove all email spam, the deduced logic reasoning would be to get rid of humans because they create the spam in the first place. Now, the means of “removing the humans” would not be a simple effort for the AI. For starters, it would need to tap into key items that could harm humans. However, before that, it would need to know what harming humans look like. Before even that, it needs to know what harm means to a human (eg. verbal vs. physical harm). All of these would be very complex “things” that it would need to learn through training data sets. And sure, someone could theoretically create a training data set for it, but then a whole slew of other questions come up. For example, if the training data set came to the conclusion that using guns to harm humans, how would the AI know how to wield one? How to use one?

Of course, that’s a more physical example. There are likely other examples that could come to fruition, like blowing up nuclear plants (see Student) or infrastructure-related damages. Overall, however, I think that as humans continue to develop AI technology the vast majority will course correct the AI towards a favorable outcome instead of a detrimental one. That’s not to say we shouldn’t be cautious and provide or create guiding mechanisms to ensure that we go down the right path.

Last question: Can AI learn common sense reasoning? I think my answer to this is that it depends on what you mean by “common sense reasoning”. The reason I say that is that common sense is in the eye of the beholder. What is logical and easy to reason through for one is illogical and complex for another.

Let’s take an example of DOTA. A group of researchers has created an AI system to play the game DOTA. This is a MOBA with 5 players versus another 5. You can choose different champions with different types of skill sets. The composition of the team heavily dictates how the game will go. After many, many hours of training the AI, it was able to beat the top-ranked players who had spent years reaching to the top. When interviewed, the players said that it had inhuman-like speeds and took completely different, unpredictable routes to beat them. The playstyle was so effective that it took the DOTA eSports field by storm with many teams adopting elements of the AI play style.

For the AI, those game mechanics were common sense. For humans, they were uncharacteristic of anything they’ve seen. So, the short answer is yes, I think in particular cases that AI can develop common-sense reasoning. However, I think that it will be largely isolated to one problem set and wouldn’t be something completely transferrable. I don’t think you could take elements of logic that the AI learned from one game and easily transfer the same sort of common logic to another game. I’m probably wrong but I’ve yet to see any studies around effective reasoning transfer between training data sets without having to go through a significant amount of model contouring, smoothing, and modifications.

That’s it for this week! I’ll likely not be writing for a week given that I’m traveling but stay tuned as there will be more content coming.

College Questions: What’s it like to run a genetics lab?

Posted by | Random Thoughts | No Comments

In this weeks post, I was asked to talk about running a genetics lab, what it takes, the challenges, how I got into it, and all the fun stuff around it. As a reminder, this series stems from conversations with college students around wild questions that they want to learn more about. I have an hour to read up on the subject and an hour to write about it. Most of the questions originate in places that I’ve thought or talked about before. This week, we’re talking about genetics and my genome sequencing facility!


I’ve written about aspects of this before however not necessarily from the business side. I can say without a doubt that it is a very different and unique domain to do a software startup in as there are a lot more pitfalls that we face.

To start out with, how exactly did we (my family) end up getting into this world? It actually all started in late 2014. My dad, who is a PhD in Biology and Bioinformatics, was working on a problem set around making human whole-genome alignments against a reference genome faster and parallel. The existing problem was that it would take hours to align a genome to a reference, causing researchers to just have to wait around forever to get results.

We talked a lot about this and discovered NVBio – a GPU-based alignment algorithm. Over the next year, we discovered that we could align 32 full 30x coverage human genomes in just 25 minutes leveraging GPUs. It was super promising! At the time, I was working at a truly big data company and we started to explore the idea of building a bioinformatics cloud. The intention was to create a hyper-performant cloud infrastructure designed just for genomics so that researchers could easily do large scale population analysis in seconds.

We spent the next year doing a ton of validation with researchers across the globe to see if the problem had legs. The good news was that the problem was definitely there, however, nearly everyone we talked too wouldn’t pay for it. Why? A few reasons.

  1. Bioinformatics stems from an academic world and there’s a lack of value perception. (“Why pay for it when I can just get it for free?”).
  2. The budgets definitely weren’t there for universities and for organizations doing genomics at this scale, they had their own infrastructure with teams to manage it.
  3. Genomics deals with massive data sets and moving that data around on public infrastructure isn’t feasible (time-wise). Upload bandwidth massively constrained this.
  4. Lastly, it doesn’t matter how good your infrastructure is…if your data sucks, you’ll get garbage results.

This last point is where the rubber met the road for us. After first-hand experience at Colorado State University and their genomics lab practices as well as interviewing many other genomics labs, pharmaceuticals, clinical researchers, and the like, we found that most produced shit quality. In some insane cases, there were labs reusing pipettes. This is a massive no-no because fragments of gDNA get caught in the micro-lining of the plastics, causing your sample to produce insane results.

Garbage data in. Garbage data out.

At the same time, while my Dad worked at the local university, there was massive in-fighting between a lot of the VPs and research leads. In one case, on VP made a big political move and, literally overnight, shut down the genomics lab my dad worked at in order to get back at another VP. My dad was part of this lab and this was the event that led us to do the lab.

The university left a lot of clients high and dry, so we decided that we were sick of the bullshit and that we were going to build our own commercial lab. We spent the next 6 months finding space, getting a purchase list in place, getting financing in place, legal work, bank accounts, website, marketing campaign, etc. This was the sprint to the new business and was extremely exhausting since we were both still holding day jobs. Apart from that, I was learning genomics on the fly. I had taken an advanced genetics course in high school but we learned and focused on Mendelian genetics. I didn’t know shit about how next-generation sequencing works, what loci were, novel alleles, drug discovery, or any of it was. I spent my evenings plowing through articles, books, software, learning from my dad, and everything in between.

Starting a genetics lab is similar in some ways to a software startup but very different in others. In the similar ways, we still had to get all the basics in place, such as accounting software, bank accounts, legal structures, a website, basic cloud infrastructure for running bioinformatics, and the like. What makes it so difficult is the wet lab portion and then integrating that into the cloud infrastructure. We call this the upstream and downstream from the genome sequencer.

The way genome sequencing works is that you typically get a sample in to the lab, such as a cell pellet full from bacteria or something. We then run it through a series of steps in which each have their own protocol. This looks like the following: DNA extraction, sample QC, library preparation, and then genome sequencing.

DNA extraction is simply extracting DNA from the sample we receive. We then do a sample quality check to look for the quantity and quality of DNA. This is done through a device called the Qubit Flurometer where we check for 260/280 and 260/230 ratios. If there’s enough to run, we then do library preparation. Library preparation is basically taking the DNA and processing it through the protocol, which consists of enzymes and other things, in order to get to tagmentation. Tagmentation is effectively tagging the basepairs with a partner that lights up under UV light in a certain color. This means that when the DNA goes through the sequencer, it can see that Blue = Adenine (as an example). This can vary quite a bit, but these protocols usually take between 8 hours to 5 full days. There’s a high degree of complexity in each of these with lots of room for error. Once the library preparation is done, we run one last quality check on a device called the Tapestation. In the most simple terms, this checks the read-length of the different fragments of DNA that are going into the sequencer. If that passes then the library is ready to be loaded into the sequencer.

Congratulations, we’ve now just performed the upstream portion of genome sequencing!

Once the sequencer is running, you have to wait 2 hours before the moment of truth comes out on whether you have a success or failed run. This moment is called “cluster density”. During the first 2 hours, the sequencer is, for a lack of better description, moving all of the DNA samples into four different quadrants. Each quadrant processes a whole slew of DNA material in parallel with each other. In our sequencer, if you have over or under 200k/mm2 by ~10%, you’re fucked. The downstream quality will end up throwing a whole bunch of quality scores below Q30. Once it’s running, you can’t stop it though and you can’t recover, so if you fuck up anywhere in the upstream process, you literally cannot recover from it on the downstream side. This means that if you have a failed run, you could be out thousands of dollars. In high-cost runs, it can be tens of thousands of dollars.

So, circling back a bit to the business side, the challenge in all of this is balancing risk, cost, and client expectations. A couple of bad runs can torpedo the company finances (and have) so it’s a very high stakes game. This was not something we had accounted for originally. Another area that was interesting in starting this up was the devices we needed. We thought there was a lot of devices that were required – and truthfully there was. However, we ended up buying a bunch of equipment that we need. As an example, we thought we would need a biosafety cabinet for sample processing. However, I think we’ve used it once? The reason being is that most people we work with send in safe samples to work with. We’re not dealing with Anthrax in our facility.

One of the biggest hidden costs of the genomics world and a large reason why there aren’t more startups in this space is that the ongoing costs are insanely high. For example, each year, just to have the basic service agreement on our sequencer is $5,000. Another crazy cost is our pipettes. These are hypersensitive in what they pull up (fluid wise) and need to be recalibrated each year. After sending ours to Germany(!), 5 weeks later we received ours back with a bill for ~$4,000. Advertising is crazy expensive because it is a scientific field. It’s not every day that someone is looking for genome sequencing so the bids on the keywords is super high, making our overall customer acquisition cost (CAC) high. Once we’ve landed a client though and they do repeat business with us, our total lifetime value (LTV) with them is very high with healthy margins. However, just getting the client and email bases is a huge journey itself.

Some of the other major challenges with this space are that each project is slightly different. Our clients are all working on insanely complicated and different research areas which typically require different ways of interacting with them. We have clients in the pharmaceutical space, clinical research, agrigenomics, and more. Even within those spaces, there are sub-verticals that we have to deal with. For example, with pharmaceuticals, we’ve done cancer sequencing, CRISPR validation, targeted sequencing, 16S metagenomics, and more. Those are all different protocols!

Having hit our 3-year mark, I can see why there aren’t more startups like ours. It truly is a very complicated field with lots of pitfalls that are extremely difficult or even unavoidable beforehand. Furthermore, doing it as self-funded, bootstrapped effort has a whole different set of problems (cash flow stabilization, net30 invoicing issues, etc.). Despite all the challenges, it’s hugely rewarding when we look at what our services help our researchers accomplish. We’ve helped accelerate research on immunotherapies for kids with rare diseases. We’ve saved companies hundreds of thousands of dollars by speeding up steps in their processes during clinical trials. We’ve helped new PhD grads gain their footing in foundational research that will help pave the way to new discoveries.

It’s all very exciting! I will admit that I never thought I’d be working on a genetics startup as that’s, by far, the last place my career or mindset was going towards. However, there’s been a lot of interesting learnings and parallels that have carried over into other areas of my life.


Alright, per the series guardrails, I’m at the hour mark of blasting my thoughts down. There’s a lot more content coming down this route so stay tuned. If you have a question that you want me to pontificate on, let me know!

College Questions: What’s going on with the USA and China?

Posted by | College Questions, Random Thoughts | No Comments

Continuing on with this series, I have a new question to answer surrounding the ever-present USA vs. China debate. The question at hand looks something like this:

Explain China, Hong Kong, and all of the global politics that are going on related to it. What is the chess game that is being played?

I feel it necessary to make a large disclaimer upfront that I’m not a pro at economics and that this is a highly, highly complex set of topics to cover. This will be pretty shallow and there will definitely be some opinions involved. I will also disclose upfront that I’m not a fan of China and tend to bias against China. As a reminder, this series is effectively me taking 1 hour to learn a subject and then 1 hour to write it – with no edits! It’s a verbal vomit of my thoughts onto the page.

Now, that all said… strap in!

To start out with, China and Hong Kong should be viewed as separate entities, as they historically were. For much of China’s history, they have been a communist state. They are still very much considered a communist state today but have a much more interesting economic agenda that we will likely cover later. In 1972, President Richard Nixon went to China with a lot of economic advisors as China was starting to undergo an economic change that focused largely on expanding their trade and presence globally.

This is separate than Hong Kong which was a British colony up until 1997 in which the British transferred “ownership” of the territory over to China proper. Since then, they have historically operated as one country but with 2 separate systems in place (communism vs. capitalism).

Why did China want to take back ownership of Hong Kong? I think the answer is pretty simple. Hong Kong had the global exposure and trade policies already in place that China could leverage. They could subtly attract global corporations to doing business with China through Hong Kong, unlocking billions of potential people to sell to. Pair that with business-friendly policies and you’ve got yourself a recipe for rapid growth. Oh, did I mention that in order to do business with China you effectively have to sign over your corporation? In essence, you have to have a company sponsor you whose origins are in China already – with the exception of a handful of large corporations (think Boeing).

This is where shit gets hairy. While China has IP rules, they don’t really follow international IP rules. For example, if you were to have your plans for your toy and manufacture them in China, you will almost certainly have a Chinese knock off to compete with your exact product at a fraction of the cost. In my opinion, the Chinese knew that Americans almost always choose the cheaper option and so they could effectively undercut American companies. The real reason they could do this is that the taxation was completely backwards. While USA corporations were not only taxed at a corporate level but also had import taxes into China, whereas China could import into the USA with effectively 0% taxes, making it impossible for USA corporations to compete. This is largely true on physical goods and gets a little more tricky with software or complex IP.

In the case of complex IP specifically with large corporations, China has been known to go to extremes to get their hands on it. In one scenario, they were actually charged for stealing IP from a massive flash memory maker called Micron.

You’d think that the USA would start to get worried about this, right? Nope. In fact, the opposite happened during the 2000’s with massive globalization. China created a massively strategic plan that was effectively a grassroots campaign to enter the USA market. In order to gain leverage on the USA, they purchased a lot of the debt we accrued during the 2000s Afghanistan and Iraq war, putting us a bit on our heels in terms of economic leverage. They then started to push massive donations into Universities to help influence agendas and normalize a Chinese culture. If you go onto any major campus in the USA today, you’ll find something called a Confucius Institute. Taken directly from Wikipedia, this is the description:

Confucius Institute is a public educational organization under the Ministry of Education of the People’s Republic of China, whose stated aim is to promote Chinese language and culture, support local Chinese teaching internationally, and facilitate cultural exchanges.

https://en.wikipedia.org/wiki/Confucius_Institute

The way I read into this is effectively a long-played game to normalize a culture that is the polar opposite to the United States.

If we fast forward to today, the Trump Administration has made it clear that they are fighting hard against China to level the playing field once again. There’s an absolute metric fuckton of political interests at play here with massive money to protect the status quo with China. While each administration has extremely shady elements going on, a great example of how deep these interests can go is exemplified in former Vice President Joe Biden and his son. Hunter Biden flew on Air Force Two to China with his dad, Joe Biden, in December of 2013. Shortly after they got back to the United States, Hunter Biden’s private equity firm received a $1.5B PE deal from China. Shady, no?

So, let’s just assume that there are a lot of deep, entrenched interests to make sure that certain agreements continue to stay in place in order to benefit a few elites. If we go along with that conspiracy theory, a lot of the follow actions make a lot more sense.

As President Trump took office, he and his economic advisors start to prepare for a trade war. They decrease taxes for the majority of Americans to start to increase the overall USA population cash flow. The trade-off here is more USA debt. From there, they start to renegotiate all of their trade deals in order to get more favorable terms. What is actually happening is that they are working to deleverage China by creating alternative trade options. He does this by creating the USMCA trade agreement, pulling out of the TPP in hopes to create a more strategic and favorable trade deal, and then starts to massive in-roads in Africa with Prosper Africa to accelerate a 4th, and cheaper, alternative labor market. The other three markets being China, Indonesia proper, and South America.

I’m sure there are many more but all of this was designed to level the playing field and deleverage China.

And then the trade war starts. It all starts to go south for China starting April 17th of 2017 after the two parties fails to make progress on negotiating on their 100-day plan. Both parties aren’t willing to make the concessions needed to move the deal forward.

What happens from there has been pretty spectacular. The Trump administration starts a “death by a thousand cuts” strategy by starting to incrementally increase tariffs on strategic items that China imports to the USA. China responds in kind by doing the same, however, the blowback to the USA is fairly marginal for two reasons: the tax cuts and the alternative markets created through the renegotiated trade deals. Or at least that is what was intended. Over most consumer goods, 9 categories of goods had seen a significant increase in CPI. Additionally, it clearly impacted the exports of agriculture from USA farmers causing the federal government to have to step in on further subsidies to keep that industry afloat.

Another element at play here is the stock market. Where money flow dramatically impacts political decisions. The Trump administration is clearly working on ways to ensure that the stock market stays “healthy” relative to China by using central financial instruments, such as quantitative easing (QE), to make sure cash is flowing and the stock prices stay high. The USA is trying to force the Chinese hands by reducing the value of their stock market while also forcing them to deplete their cash reserves. The problem with this strategy is that China is considered a currency manipulator. This makes it very difficult to compete on cost of goods imported/exported. For a quick primer on currency manipulation, check out this article.

Alright, so we’ve talked through a bit of the trade war. How does this deal with the Hong Kong uprising?

The reason the Hong Kong protests started to happen was due to something called the ELAB: the Anti-Extradition Law Amendment Bill. Hong Kong citizens oppose this bill because it would enable China to detain and extradite Hong Kong citizens back to China, effectively saying that China rules supersede Hong Kong. This is scary for Hong Kong citizens because China has a very Orwellian approach to governing its citizens. Millions of citizens have been part of this demonstration.

The big reason there is such a blowback is that Hong Kong has been largely pushing for a more democratic society – similar to the USA. There was an attempt to start the revolution back in 2014 with the Umbrella Revolution – a series of sit-ins by students to protest the Chinese government starting to take over the Hong Kong system. There is effectively a multi-government war going on and the Hong Kong citizens are being subjected to being taken over into communism.

As such, you can see the Chinese government continuing to push for a takeover in order to gain a strategic economic lever inside of Hong Kong. It wouldn’t surprise me if the USA government was trying to stir the pot with the local citizens as well.

I’m at the 1-hour mark so I’ll do a quick summary to wrap this up.

  • The USA renegotiated strategic trade partnerships to gain more favorable terms to deleverage China
  • The USA is starting strategic investments into Africa to further open up more markets to buy/sell into in order to deleverage China
  • The USA is incrementally increasing tariffs on China in order to push its stock market down while starving their government tax revenue
  • The USA is betting on the financial “squeeze” to push China to negotiate a fair, bilateral trade agreement as well as renegotiate US debt held by China
  • China has a long term strategy of normalizing their way of operations by effectively infiltrating the USA from the ground up (eg. University influence on the new generation of kids)
  • The Hong Kong protests pose another angle of a challenge for China that is making it hard for them to continue down the path they’re on.

I wouldn’t be doing my job if I didn’t call out one specific example that hits home for me. My genetics company recently had a call with a very large buyer of genetic consumables for genome sequencing from a company called Illumina. In our discussion, we questioned why there are such few organizations in the USA doing whole human genome sequencing. Their answer was astonishing.

What has effectively happened is that a company called Novogene, who is owned by the government of China, created entities in the USA to operate within. The Chinese government buys massive quantities of consumables at a super high discount rate. From there, the Chinese government subsidizes the cost of goods, labor, and real estate in order to bring the price significantly down to a competitive range that USA companies cannot compete with. Companies do business with Novogene because it’s the cheapest option on the market for whole human genome sequencing. Typically, human WGS costs around $800-$1,000 to perform. With Novogene, it’s about $400-$500. What is so bad about this is that the samples are physically shipped to China where the genetic data produced lives. This data is critical for novel allele discovery when aggregated. Now, obviously you need health records for truly targeted drug discovery, however, for a wide range of generics, you can work off of population-based genetic data sets. The generics drug market is on track to reach $380 billion by 2021.

For us little guys, it makes it nearly impossible to compete with the entire backing of the Chinese government versus our measely $200k investment.


I hope this was interesting to read. Leave a comment if you agree, disagree, have more comments to add, or whatever. I’ll be continuing to play around with this series so if you have a question that you want to me to pour my brain over, let me know!