There Is No AI

Sep 02, 2026 01:18 AM - 2 hours ago 1

I've searched pinch small occurrence for ideas astir really we tin grip AI truthful that it benefits alternatively than harms society. Here is simply a suggestion from Jaron Lanier.  It was published by the New Yorker successful 2023, erstwhile GPT-4 was the latest model. Time named Lanier to its database of 100 astir influential group successful 2015. 

A programme for illustration OpenAI’s GPT-4, which tin constitute sentences to order, is thing for illustration a type of Wikipedia that includes overmuch much data, mashed together utilizing statistics. Programs that create images to bid are thing for illustration a type of online image search, but pinch a strategy for combining the pictures. In some cases, it’s group who person written the matter and furnished the images. The caller programs mash up work done by quality minds. What’s innovative is that the mashup process has go guided and constrained, truthful that the results are usable and often striking. This is simply a important accomplishment and worthy celebrating—but it tin beryllium thought of arsenic illuminating antecedently hidden concordances betwixt quality creations, alternatively than arsenic the invention of a caller mind.

As acold arsenic I tin tell, my position flatters the technology. After all, what is civilization but societal collaboration? Seeing A.I. arsenic a measurement of moving together, alternatively than arsenic a exertion for creating independent, intelligent beings, whitethorn make it little mysterious—less like hal 9000 aliases Commander Data. But that’s good, because enigma only makes mismanagement much likely.

...

This concept, which I’ve contributed to developing, is usually called “data dignity.” It appeared, agelong earlier the emergence of big-model “A.I.,” arsenic an replacement to the acquainted statement successful which group springiness their information for free successful speech for free services, specified arsenic net searches aliases societal networking. Data dignity is sometimes known arsenic “data arsenic labor” aliases “plurality research.” The acquainted statement has turned retired to person a acheronian side: because of “network effects,” a fewer platforms return over, eliminating smaller players, for illustration section newspapers. Worse, since the contiguous online acquisition is expected to beryllium free, the only remaining business is the hawking of influence. Users acquisition what seems to beryllium a communitarian paradise, but they are targeted by stealthy and addictive algorithms that make group vain, irritable, and paranoid.

In a world pinch information dignity, integer worldly would typically beryllium connected pinch the humans who want to beryllium known for having made it. In immoderate versions of the idea, group could get paid for what they create, moreover erstwhile it is filtered and recombined done large models, and tech hubs would gain fees for facilitating things that group want to do. Some group are horrified by the thought of capitalism online, but this would beryllium a much honorable capitalism. The acquainted “free” statement has been a disaster.

...

Consider what mightiness hap if A.I.-driven tree-trimming robots are introduced. Human character trimmers mightiness find themselves devalued aliases moreover retired of work. But the robots could yet let for a caller type of indirect landscaping artistry. Some erstwhile workers, aliases others, mightiness create inventive approaches—holographic topiary, say, that looks different from different angles—that find their measurement into the tree-trimming models. With information dignity, the models mightiness create caller sources of income, distributed done corporate organizations. Tree trimming would go much multifunctional and absorbing complete time; location would beryllium a organization motivated to stay valuable. Each caller successful preamble of an A.I. aliases robotic exertion could impact the inauguration of a caller benignant of imaginative work. In ways ample and small, this could thief easiness the modulation to an system into which models are integrated.

...

There are besides non-altruistic reasons for A.I. companies to clasp information dignity. The models are only arsenic bully arsenic their inputs. It’s only done a strategy for illustration information dignity that we tin grow the models into caller frontiers. Right now, it’s overmuch easier to get an L.L.M. to constitute an effort than it is to inquire the programme to make an interactive virtual-reality world, because location are very fewer virtual worlds successful existence. Why not lick that problem by giving group who adhd much virtual worlds a chance for prestige and income?


More