海角社区官网

The creators of other content I wanted to include were happy to share; others gave me a flat no.

But nobody from Amazon, Google, Microsoft, Meta or any other big-tech firm is likely to ask me or the publisher of this news organization for permission when articles like this are poured into the gaping maw of their artificial intelligence machines.

Figuring it鈥檚 easier to ask forgiveness than permission, most of these dominant players are helping themselves to the work of writers, photographers, musicians, artists and creators of all kinds to feed and train hungry AI beasts.

And there鈥檚 nothing the beast likes better than dependable and creative material from reputable human sources.

Anybody who has used ChatGPT, for example, knows it is about as reliable as schoolyard gossip.

But large language models (LLMs) are learning quickly, using material collected from news agencies like this one, or from books and artists and photographers and anything else they can get their hands on to learn and improve the material they create聽鈥� the poems, the essays, the summaries, the stories they write.

Accurate, well-researched, fact-checked, foot-noted, ethically sourced and creatively written material聽鈥� by humans聽鈥� is not just an elixir to LLMs, it is lifeblood. They can鈥檛 succeed without it.

So the race is on among big-tech companies to feed LLMs as much of this valuable information as possible聽鈥� without permission or payment if necessary聽鈥� to help generative artificial intelligence get more intelligent, and make untold fortunes in the process, leaving human creators, as usual, holding an empty bag.

Already, they鈥檝e used billions of articles, webpages, videos and codebases of copyrighted material to train LLMs.

Payment? Permission? For these big players, there鈥檚 simply no time for such nuisances. It鈥檚 little more than an inconvenient question from poor artists and beset news organizations.

Eric Schmidt, the former Google chief executive and chairman, recently gave students at Stanford University an idea of how big-tech billionaires think: 鈥淚f TikTok is banned, here鈥檚 what I propose each and every one of you do. Say to your LLM the following: 鈥楳ake me a copy of TikTok, steal all the users, steal all the music, put my preferences in it, produce this program in the next 30 seconds, release it, and in one hour, if it鈥檚 not viral, do something different along the same lines.鈥� That鈥檚 the command. Boom, boom, boom, boom.鈥�

The idea, he explained, later telling students not to quote him, was to loot the material now and worry about the consequences later. In other words, hire some lawyers with the money you make and let them clean up the mess in the fullness of time.

Others have insisted that Amazon, Google, Meta, Microsoft and others are not stealing this material at all; it鈥檚 simply fair use (or in Canada, 鈥渇air dealing鈥�). That is preposterous.

What exactly is fair use? As any author or book publisher knows, limited, careful use of copyrighted material for education, criticism and more is a grey area.

At the moment, quoting with attribution one hundred words in a 100,000-word book might be considered fair use; but then again, it doesn鈥檛 mean you can鈥檛 be sued for copyright infringement. For example, I wanted to use 200 words from the famous 鈥渂lue sweater鈥� scene in the movie “The Devil Wears Prada” for my book “Shopomania,” but I wasn鈥檛 about to irk the giant Disney corporation without checking with it first.

It鈥檚 quite clear that LLMs are slurping up more聽鈥斅燼 lot more聽鈥� than just that. Social media and the internet have made copyright infringement even greyer, but what about gobbling up the whole book, digesting it quickly and then spitting out content based on what it just absorbed? How can that be fair use?

There鈥檚 nothing 鈥渇air鈥� about tech monopolies helping themselves to published content聽鈥� all in the name of profit and market dominance聽鈥� when it鈥檚 obvious licensing is a more sustainable solution.

These big tech firms聽鈥� some of the largest companies in the world聽鈥� have tens of billions in revenue and investments. And there may be trillions to be had.

Book publishers, meanwhile, are struggling like never before. And it is the rare author or novelist indeed who makes a living wage at a lonely keyboard, let alone earning enough to take legal action.

News organizations face well-documented challenges in an internet world. Forced to be dependent on digital giants to reach audiences, distribute and monetize content, the theft of their material is only one of many obstacles to survival.

Newsrooms are closing everywhere. Journalists are leaving the industry. Many rural, small and even mid-sized communities are becoming news deserts. Some people wrongly think journalism is cheap to produce.

And misinformation, increasingly spread on big tech鈥檚 platforms, is polluting society.

Some publishers have demanded AI companies stop scanning their websites. Others have charged them with copyright infringement. A few have made deals.

But it鈥檚 not enough to simply enforce copyright legislation or arrange licensing. Other tools are needed: taxing AI companies and subsidizing creators; using anti-trust laws to prevent abusive dominance; and breaking up monopolies to create even playing fields.

Finally, should these platforms be treated as essential聽鈥� that is, as common carriers with public obligations beyond their own bottom lines?

The Writer鈥檚 Union of Canada has signed on to a statement by more than 30,000 creators: 鈥淭he unlicensed use of creative works for training generative AI is a major, unjust, threat to the livelihoods of people behind those works, and must not be permitted.鈥�

In the absence of a clear signal from industry, the courts and government that mass theft is against the law, this procrastination and a hodgepodge of fixes will benefit the big players and cost the small ones until it鈥檚 too late.

Artificial intelligence technologies are a fact of modern life. If they were ever regarded as benefiting society as a whole, they are now a way to make billions, perhaps trillions, for a very few.

Like all technology, it may well harm us聽鈥� or, as some AI pioneers insist, perhaps even destroy humanity. It may also help us adapt, innovate and improve. But if that latter scenario comes to pass: how, and for whom?

Creators can鈥檛 simply ignore it, but we cannot let tech giants decide the future without consulting with the rest of society.

We are left in a situation in which AI has left us in a series of impossible binds.

Bad information has never been more pervasive. Good information has never been more valuable.

Bad information is costing lives across the planet. Good information is a key to a peaceful and equitable future.

Bad information is worthless. Good information should not come cheap.

The tension between these things cannot be left unresolved. We must confront the threat AI poses to creativity and information, lest the human work that sustains those things is also erased from history.