So, if you were at a stage of your career where you're trying to find interesting fields and industries that might see huge growth, where would you look?
I landed myself in a shop that sells, essentially, tiny locally-networked systems of specialized Linux boxes. Yes, our core offering is running on the JVM, but all along the edges there is just so much Bash and PHP.
I don't think I'll ever work in a place again where I have this much opportunity to become a genuine old school shell and webshell wizard again. I want to master the Primordial Arts, their endless exceptions to exceptions, and come out the other side as a true master of something completely fucking ridiculous.
I did it once with Microsoft PowerShell. I can do it again.
Pay is probably under 50k a year though. Can't win em all.
I'd love to hear more about the product though! Without giving too much away, what does the product/company do? :)
[0]https://www.zyvexlabs.com/apm/products/zyvex-litho-1/
Lot of interesting things happening with NeRF, Gaussian splatting, and similar. We're close to be able to do full 3D video capture of locations and high-fidelity playback/exploration in VR.
I'm so glad that some companies/people are building and releasing models to the public. This amazing development being exclusive to OpenAI would've been a terrible situation for the world.
Academia has become a game and tech development has become driven by pure consumerism in the race to replace human relationships with the experience of "The Product" so that we have to continually upgrade to maintain any semblance of social cohesion.
There is very little left in tech development that is good. We should think about being more sustainable instead.
Tech not invented but popular in sci-fi like transporters or replicators could end hunger or change how we move. 3d printing has reach a general price point on par with other manufacturing but when it does it could change things.
Plenty left that is good and necessary. Don't give up because llms and social networks are overhyped.
One of my issues as someone who works in tech is that it so much capital has poured in and so much attention was on the field that there are relatively few innovations where one can create value. OpenAI being an exception that also proves the rule. A true innovation that literally needs billions of $ to have a chance of happening. Microsoft or Amazon didn't need that kind of money to build successful companies on the back of new business models.
The same incentives that enable google to exist, then push the same company to be "less good" by prioritizing growth and public market interests.
As long as you borrow capital, there is always a cost on incentives.
Consider the DVR. It transformed the experience of watching television before cable companies figured out they needed to acquire it and worsen it. The damage (to them) is done, we got used to fast forwarding through commercials and now that's a necessary feature with cable.
That could never happen today. Profit, advertising, and surveillance are too strongly built into the process of invention right now. Technology companies have mastered enshittification on a level that older industries never innovated on their own.
This isn't proof that it won't happen again, it just says that the rate at which new innovations in software (and a new workout app doesn't count) has slowed down. LLMs being the exception but it's not like you and I can start an AI company in our garage.
Create decentralized software and get ahead for the next shift.
My best bet is medical tech and especially body modification and enhancement. While I have been excited and unafraid of all new inventions my whole life, this is a topic that makes me more uncomfortable than excited - a reaction I often observed in people older then me with other technologies. So I'd say invasive body enhancements and augmentation is a good candidate to be the next big thing.
[1] There are always exceptions, but the overall picture is pretty obvious.
Software is an enabler - when it becomes the focus, I think something has gone massively awry
Upscale those laser zappers which are now being fielded to combat drones and put them in high-flying aircraft or on mountaintops. Start zapping the debris and clean out those orbits. Larger pieces of junk can be collected by some iteration of MegaMaid™ (in essence an orbiting Roomba). Problem maybe not directly solved but at least postponed to be either solved this way in a number of years or naturally through orbital decay of circulating junk.
> The Economics of Orbit Use: Open Access, External Costs, and Runaway Debris Growth (Rao, Rondina, upcoming 2024).
https://www.journals.uchicago.edu/doi/abs/10.1086/730695
And for an arvix/working paper link: https://arxiv.org/pdf/2202.07442v2.pdf
> In our main calibration, Kessler Syndrome can emerge anytime between the year 2040 and the year 2184, with the precise date being very sensitive to the calibration of autocatalytic debris growth parameters.
https://news.ycombinator.com/item?id=37473795 (Ask HN: Tech that seems to have vanished?, 189 comments)
Memristors, alcohol batteries, laptop frame heatpipes, flywheel energy, smart textiles, and several others were mentioned.
1. Become an expert in using and augmenting AI tools to accomplish your work, whatever that may be. There is a difference between average and expert in the results that can be obtained. AI expertise will be important at least for the next 10 years.
2. Commit to lifelong learning and adaptation. The constant for the remainder of this century will be continuous, exponential change. When change has a slope near vertical, this should be apparent.
3. The closer you can be to hard science and technology in education and work, I believe the better chance you will have to make significant professional contributions. The CEO of Nvidia said as much in a recent interview.
4. Make a plan by looking at every professional career option and ask this question: "what is the likely result for this field when AGI becomes real and commonplace?" Then choose to examine more closely those fields where you believe you can thrive in the midst of change.
I hope this is helpful, and enjoy the journey!
I'd love to do hard science, rather than my current meaningless consumer webapp job. But it'd mean... giving up 5-6 years of tech salary to go back and get a phD, and then post-phD my job prospects would be geographically limited to a couple locations where all the science/research jobs are, with worse hours, worse flexibility, less stability and drastically less pay.
It's hard to say "I should quit my fully-remote, wfh, $200k, low-stress, unlimited PTO software job to work in a lab in Boston for 50 hours a week for $90k". I don't have any solution here, kinda just venting through my golden handcuffs.
I never got this frame of mind. You chose to wear them. You like your golden bracelets. You can take them off any time you want. You are making the choice of staying. At the very least assume the responsibility.
>You chose to wear them. Correct. Because it is the least-bad option that I see as a possibility.
>You like your golden bracelets I do not. I dislike my handcuffs. I just recognize that the other handcuffs I'm allowed to replace them with are worse.
>You can take them off any time you want. You are making the choice of staying I can replace them with different, worse handcuffs. I'm unhappy about being in handcuffs, but if someone suggests I replace my golden ones with the sharp, rusty ones, I will decline and choose to stay in my current ones.
>At the very least assume the responsibility I am? In fact the degree of agency here is central to the whole point. If you tell a prisoner they can choose between death by lethal injection or death by being burned at the stake, they can make the choice of lethal injection and "assume the responsibility" for that, but it's not like it's what they actually want-want. It's what they want under the constraints of the circumstances.
I wish I had paid more attention in maths class when I was college tho, but it's never too late.
AI sure seems like a good bet. I am also keen on what other fields are out there. I've been trying to understand what are the areas where there's something brewing and will likely breakout in the next 5-10 years.
If that sounds too abstract: less safetyism. We could see new kind of vehicles, sports and space exploration.
So long as we ditch "safetyism" - OSHA, et al, have their place ... but they have become massive bureaucratic nightmares no one can fully "comply" with, and whose regulations are a confusing mishmash of contradictory 'advice'
Space travel to Mars and space colonization/mining.
Brain-computer interfaces.
If you are "Posix compatible" plus 'This Other Thing' ... you have the makings of a great platform
Some of the end goals could be: (1) program verification becomes something that's done by default, (2) the entire mathematics literature becomes formalized and verified (and all errors there found and corrected), with programs able to read and formalize papers from the old literature, and (3) new mathematics is done largely by computer, with people acting as "is this interesting?" filters more than agents of discovery. I don't see any reason why mathematics should be immune to domination by computers, any more than (say) Go (the game) was.
Inshallah. I hope you're right.
> interesting fields and industries that might see huge growth
The porn industry
Honestly ... I think the rise of [almost breaking the uncanny chasm] AI-generated images and video is going to both help and hurt the porn industry
How will it 'help'? Generate, on-demand, whatever form of sexual fantasy the user can prompt
How will it 'hurt'? Generate, on-demand, whatever form of sexual fantasy the user can prompt
When actual, IRL humans are involved, there are always limits to what can be produced
When that limitation is gone ... the cuffs will be off!
Is AI-generated 'underage' sex still "pedophilia"? When those 'engaging' in whatever activity the prompter has suggested 'do not `exist`'?
What about so-called 'extreme' BDSM/pain/fetish/etc? If the 'actors' are all AI-generated ... is it still [potentially] `wrong`?
There are a HOST of ethical and legal issues that are about to hit the porn industry ... and those who want to - rightfully - prosecute those who want grossly-illegal/-unethical material!
Strictly by the letter of the law ... it 'cannot' be `kiddie porn` if the images are not of real people.
Yet ... by the spirit of the law ... it most certainly is `kiddie porn` if the images are of obviously `underage` individuals!
Or what about so-called 'snuff videos'? It is actually a beheading if the `beheaded` is AI-generated?
The legal system is going to have a LOT of work in front of itself in this dawning era, to be sure!
I wouldn't look for a specific industry or anything interesting. It's all pretty much corporate BS. Take whatever job pays pretty well and isn't obscure tech so you'll be able to switch jobs. Finance, law, and institutionl sales (gov, schools, hospitals) aren't going anywhere anytime soon.
Is "kids hooking up with each other?"
I imagine a day when it is socially unacceptable to hire someone to do something for you. You have AI and Amazon, why not learn, order the parts, and do it yourself? Imagine a world where we all work a few hours a week but occupy ourselves learning new skills and applying those skills to improve our environments. This would also keep us ready to be useful at work.
I am reminded of a famous quote: "A human being should be able to change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly. Specialization is for insects." -Robert A. Heinlein
Model X trained on all our data thinks Y is ok and that permeates lessons here and there. Raising a generation that does not rightfully challenge Y as it should.
As a somewhat stupid example: had LLMs been invented in thr 1800's would they teach whatever topic with an underlying sentiment that slavery is a just and natural state?
Teachers and tutors have always been an influence in the formative years of younger minds. I am afraid of substituting that human influence bringing intellectual and moral challenges for a one-size-fits-all frozen in time pasteurized worldview.
To be clear: LLMs as educational tools; inevitable. LLMs as teachers - no, please
Let me know if this is a stupid feeling (and a feeling is all I have) or if it is too much of a Old Man Yells at Cloud thing
It is likely that a main LLM will dominate the play field (network effects) and it will breed a thinking monoculture. The more entrenched it becomes, the less resilient society will be, because we will become over reliant on its model weights.