peace with the thinking machine
What would happen if we lived with an ai that wasn’t evil? Despite the headlines it doesn’t must be an apocalypse technological progress even when destructive rarely goes back in the bottle, but the difference between great achievements and total chaos is how we use it. The concerns involve utility usage, copyright and patent law, worker displacement especially for early office careers. Land is built and there is noise. Most can be regulated or engineered around.
environment
The nightmare of all water being consumed by it is a hyperbole even at projected growth they will unlikely out use golf courses anytime soon. You don’t see nearly as many calls to ban golf. The water usage concern is about evaporative cooling and failure to recycle. There are countries building extremely low usage models sunk on the continental shelf somewhere. issue there becomes underwater maintenance and these are not going in pristine ecosystems; the main test sites are off major port hubs. The main usage of water in a data center is to cool components in a spray or evaporative server cooling. Is recycling possible? or use of other cooling? Yes, theoretically, but the purification to avoid scale or corrosion risks costs money so there needs to be an externality on the water use factored in. To encourage recycling, it must therefore be more expensive than freshwater withdraws. And the regulatory frameworks need to work and update. The infrastructure may be if water cooled, which is the most workable economically and engineering wise, may not be available everywhere. Climate change doesn’t help with this issue. Even zoning this activity into humid regions? and municipal utilities do need to be aware of predicted use in a good faith estimate before signing the approval. It may not drink everything on the continent, but it can cause local utility strain particularly on smaller grids or fluctuating or smaller water bodies. Noise issues a sound wall or several feet of vegetation between center and residential homes can solve it in most areas where there is preexisting background noise. And energy they should be factored in as a major limiting factor like water, but the genie isn’t going back in the bottle.
Capabilities and job risks
The other concern for work is it can compete with some things but as of yet lacks hands and between hallucination risk and poor judgement (try asking it for life advice or a challenge) which limits its intrusion and art generated by it does need copyright law reform. but genres like social media, internet, even electricity and radio are unlikely to go back into the bottle. Pretending otherwise prevents solving anything or addressing it. And the construction crews are fond of large, long–term projects; it may not be a permanent major employer, but construction and maintenance and updating will employ many for whom a 2–3–year contract is stable. Construction measures career and work stability differently from job sites change but work in the region and becoming increasingly specialized and therefore valuable is a major part of career progression in trades. Particularly if not ambitious of management on site. it’s why they have held out as more unionized than most sectors. Stability is provided by projects, especially industrial and institutional ones in the region, and long commutes are common. a 2–3-year job is a long stable stretch one a renter can move closer for a better commute potentially.
The longer-term job risks are still uncertain with administrative and knowledge to work most vulnerable along with some service sector. Other areas will depend more on human hands, at least until the machines get mobility and fine motor skills. Forecasts range from the cautious who suspect it will augment current work and create role shifts mostly to signs it will turn us into a more industrial and physical work workforce a few enthusiasts many deeply invested in it are claiming eventual mass layoffs. This is a case for unionization and trade reskilling. More theoretically, in another essay it could be if labor had sufficient power support measures such as a shorter workweek or allow for greater productivity. And an important reminder of a similar thing happened to the trades and manufacturing in the automation boom of the late 20th century. The solution has not been to remove the innovation people shifted to service and knowledge work granted the bigger issue was a simultaneous loss of union membership and labor power. Technology is neither good nor bad for the employee inevitably an employer will generally attempt to minimize labor standards this has been true since feudalism it only improves through state action and collective bargaining the horrors of the industrial revolution were often in at least some way an improvement over preindustrial society. The benefits to the bulk of the population and things driving the billions out of abject poverty over a century is a combination of productivity gains and a population with the power to ensure those gains at least partially reach them. Through reform and organization
Creatives and intellectual property
Get thoughts of comentary’s stories in your inbox
Join Medium for free to get updates from this writer.
creativity and it are also factors that can the machine create images or write something articulate and potentially cause flooding yes are their things it does poorly yes and is creating with it a skill arguably. It can in a few prompts create an image that can match casual observations but rarely looks all the way right if attempting realism and usually not beautiful without signs of effort on a human creating either takes knowing how to ask and refine inputs is a process of its own not a true art but a tool a digital medium almost. It can be a tool or medium. And it is a question of copyrights that does need legal questioning to decide who gets rights to what. Prompters should get a claim, but if it can substantively be traced to an identifiable human made piece, it needs basically a bibliography code on the image. And the tool should be declared though it often has hallmarks. Especially low effort pieces.
Preexisting trends
it raises fundamental questions but can be used in many ways from the silly assistant to the confidant (even if a lousy therapist that requires human judgement and delicacy) to the artistic medium to instructions it does recipes and craft directions well. And most economic precarity it is blamed for can more accurately be traced to the pandemic and the recent inflection, the 2008 recession which much of America never fully recovered from, or the deficit and inequality skyrocket created by the Nixon and Regan administrations move to a lower and less progressive tax system and reduced social, public, and R&D spending coupled with unions being undermined. reaching a crisis point during the pandemic and in the recovery, shock followed by political shocks. Still the job losses exist but most of it is a policy the nature of work may shift but if workers had a bargaining ability and we had a functioning democratic system it could be responded to proactivity is key here unfortunately it is a struggle of our institutions. The same technologies finding major uses in organization, math, science, and communications can also be a tool to survive, and can’t be the final authority.
Do we want them here on our terms or abroad on someone else who may not be a geopolitical friend the Chinese are competing with us on this and we have current allies that are being alienated and excellent candidates to build the infrastructure
conclusion
peace with ai is possible but requires proactivity and fights for use and management that works for society and major corporate control plus a major shift the end of Reaganism we need the labor protections and safety nets of the great society, education and retraining on a massive scale, strong jobs and public works programs, public research with guardrails. And strong democratically accountable systems that somehow can get things done. None of which is carried out by imitating an ostrich.
China Powers AI Boom with Undersea Data Centers | Scientific American
