in February, Trump signed a five-point plan to assure AI makes America Great Again, down the line. The five: to promote R&D, reduce barriers, train workers, foster trust, and compete globally, really amount to one:
(b) The United States must drive development of appropriate technical standards and reduce barriers to the safe testing and deployment of AI technologies in order to enable the creation of new AI-related industries and the adoption of AI by today's industries. [i]
In other words, hand the government's knowledge, resources, and labor over to private companies.
In their usual, contradictory way, the gatekeepers of capitalism - your officials - mandate, the future is free-market-based Ai. Of course, this is just to house-keep. The plan was already working. Venture capital funding for AI startups reached record levels in 2018, increasing 72% compared to 2017 to $9.33bn in funding. [ii]
Digital technologies, in general, appeal to investors. In part, because they forego a lot of the hard inputs of manufacturing. In part, because the workforce is young and still willing to compete against itself. And in part, particularly with AI, because we don't really know what it does yet, they can promise anything.
AI has particular-value to speculators, as it makes analysis of highly-complex data more within reach. Yet it's important to recall, markets themselves are a form of predictive technology, and speculators have never really minded that their predictions and outcomes rarely coincide, so long as they further casino capitalism.
Thus, besides what additional risks AI brings, it inherits all the problems of the market. Namely, that its predictions are only a synthesis of its inputs. When they fail, as they do, we invert blame to 'market-failures', rather than the fruition of a biased set.
According to a recent New York University research center report (published by the AI Now Institute, as one of 150+ studies with like conclusions), AI is adopting our hierarchies and prejudices. The industry's overwhelmingly white-male workforce has gifted us systems that perpetuate race and gender biases. [iii]
For instance, in 2016, Pro Publica ran an analysis on a proprietary algorithm, COMPAS, that predicts recidivism rates to weigh in sentencing. It "found that black defendants were far more likely than white defendants to be incorrectly judged to be at a higher risk of recidivism, while white defendants were more likely than black defendants to be incorrectly flagged as low risk. " At its best it proved 60% accurate. Among other miscalculations, it was incorrect in 80% of the cases regarding violent recidivism. [iv]
Of course, (if you're not Pro Publica) the role of our so-called free press is to glean otherwise critical studies for their nod toward market-based solutions. In this age of fashionable, collective self-criticism, its sexy to endite the status quo (the market-mandate), but it rarely if ever takes outside the bounds of market-capitalism.
In this case, the report suggests making pay-rates, harassment, and discrimination reports public, and increasing affirmative action to dilute AI's prejudices. [v] In other words, these are market-failures. To which they prescribe market tweaks.
Add to that, Trump's mandate gets frequent criticism for not adding a sixth point facilitating immigrant contributions to R&D. This, from both sides of the isle, not as politics, certainly not to preach human rights, but as a necessary market-adjustment.
In sum, these are market corrections to address market-failures. Already there's a contradiction (dating much-further back the AI). Simply put, you cannot tweak your way out of failure.
Today, market-failure on a grand-scale may end capitalism - and civilization - for good, through climate-change. How confident should we be that market-corrections will avert full-scale disaster? The common fear is that they'll produce too little change, too late.That's not lost on investors. Consider the insurance industry has openly acknowledged climate impacts for 30 years. However, they remain the largest American investor in fossil fuels. Their response has been to increase rates according to risk, and in a few cases withdraw from the market, if the chance of payout was too great. (Cases in New Orleans and Atlantic City are examples.) A recent Brookings Institute paper, How artificial intelligence will affect the future of energy and climate, concedes: AI will be a central feature of climate-science, but a hyper-smart, AI-driven energy system still won't net the 60 to 100 percent emission reduction needed to halt global warming. [vi]
But to say it's not up to the task already misleads us. We should not assume markets, and that includes any market-based technology, will help, whatsoever. Rather, technology has equal ability to tighten old controls, as to prove the efficacy of new ones. (Recall, for example, the very term 'climate change' is a market-correction of the more informative, 'global warming'.) Common sense says, if it relies on the market for its development, defending the market will be priority, even if that means killing it in the long run.
To quote the Brookings report:
AI helps make markets more efficient and easier for analysts and market participants to understand highly complex phenomena from the behavior of electrical power grids to climate change. The question is whether there is a "bias" in how AI-related technologies affect energy supply, such as whether they're making traditional hydrocarbon suppliers more productive faster than they make zero-carbon renewables more productive. [vii]
The author notes, for example, AI is particularly well-suited for mapping complex underground reservoirs and tailoring drilling methods required to mine oil and gas from shale. That may be true, it's better at that than managing wind farms. It's far outside my area of expertise to say. But I'd wager, the preference has more to do with the bent of its patrons. And obviously, it's going to do the task we set it to.
In other words, as with race and gender issues, as with prison sentencing, so with climate change, we've primed AI to pick the historic winners, not the rightful ones.
Mind, the paper still found the glass to be half-full (at least for the market).
Digital technology lets companies amend their scale, focus, or their cargo on short notice. Thus, it incentivizes reacting to the market, rather than building it in the Fordist sense. AI is, itself 'adaptive technology'. Ergo, AI prevails at 'adaptation-strategies'. But if that's the model - predictability, instead of prevention - climate policy will preference "adapting to climate impacts and implementing quick responses in case of climate emergencies", rather than taking the broader steps to avert them.
Of course, how we develop AI, only matters if we also implement it. That requires a political program, outside of development. Here it should be revealed why a made-for-TV type swindler like Trump chimes in on something big like AI. The goal, we said was to 'reduce barriers.
Notably, the (Trump) initiative opts for a market-based approach to regulate AI. This is a good strategy. "AI could help reduce one of the greatest dangers as societies develop adaptation strategies, which is that they commit vast resources to adaptation without guiding resources to their greatest value. High levels of uncertainty, along with acute private incentives that can mis-allocate resources for example, organized labor might favor some kinds of adaptive responses (e.g., building hardened infrastructure) even when other less costly options are available. [viii]
The report concludes, highly-adaptive technologies could help "tamp down enthusiasm for regulation and make practical a greater reliance on market-based instruments such as carbon taxes". (Emphases mine.)
So, is AI quickly becoming capitalism's rear guard? If yes, why are fantasizing about our dystopian future, when it has the dystopian present in mind?
Where are the Luddites?