Understanding machine learning (ML), generative pre-trained transformers (GPT), and large language models (LLM) has become a part-time job for me. Not only is there a lot of information and discussion, but a wide range of opinions. The topic of ‘AI’ constantly pops up in professional meetings. Researcher danah boyd discusses the difference between the perspectives of automation vs. augmentation as ‘AI’ develops.
“When it comes to AI’s potential future impact on jobs, Camp Automation tends to jump to the conclusion that most jobs will be automated away into oblivion … most in Camp Automation tend to panic and refuse to engage with how their views might intersect with late-stage capitalism, structural inequality, xenophobia, and political polarization … Camp Augmentation is more focused on how things will just change. If we take Camp Augmentation’s stance, the next question is: what changes should we interrogate more deeply?” —Zephoria 2023-04-21
I am mostly in the augmentation camp, though I am concerned that automation + capitalism = a perfect storm. This was the case with the augmented work enabled by the personal computer. Knowledge work improved significantly but wages did not.
“The fact that personal computers didn’t raise the median income is particularly relevant when thinking about the possible benefits of AI. It’s often suggested that researchers should focus on ways that A.I. can increase individual workers’ productivity rather than replace them; this is referred to as the augmentation path, as opposed to the automation path. That’s a worthy goal, but, by itself, it won’t improve people’s economic fortunes. The productivity software that ran on personal computers was a perfect example of augmentation rather than automation: word-processing programs replaced typewriters rather than typists, and spreadsheet programs replaced paper spreadsheets rather than accountants. But the increased personal productivity brought about by the personal computer wasn’t matched by an increased standard of living.” —The New Yorker 2023-05-04
These are my observations and conclusions to date, which of course may change in a month or two, given the speed of technology development in this field.
- It will become much more difficult to be an average — or even above average — writer. Only the best will flourish. —2022-12-13
- At this time we do not need more ChatGPT experts. What we need are people who can see these technologies in the context of human work. We need more ‘what-if’ questions to be asked. —2023-01-21
- Human oversight of machines and software will be essential as these systems permeate our economy and society. —2023-02-27
- GPT extends our voices, obsolesces many forms of human writing, retrieves the ancient Oracle of Delphi, and when pushed to its limits reverses into Potemkin villages. GPT provides instant synthesis that may be good enough, but often contains errors. If it gets used to answer all our questions then it becomes a tool for “aphoristic nincompoops posing as techno-oracles”. —2023-03-15
- What happens in several years when workers have not had to ‘manually’ develop writing skills? If they have always started with the output of a GPT or LLM tool, will they develop good basic writing skills? —2023-03-19
- “So what happens when we automate our most impactful and superior cognitive capacity—thinking—and we don’t think for ourselves? I think we end up not acting in very smart ways, and then the algorithms are trained by behaviors that have very little to do with intelligence.” —Tomas Chamorro-Premusic—2023-03-26
- The emerging ‘AI’ is creating a situation of, “the total decoding and synthesizing of reality.” Everything we do runs on language — laws, friendships, relationships — and now non-humans can create persuasive narratives. —2023-04-11
- The machines are our tools, not our friends. Let’s make sure we know how they work and what is behind them by learning with and from each other. —2023-04-18
- “Tech companies have grown secretive about what they feed the AI. So The Washington Post set out to analyze one of these data sets to fully reveal the types of proprietary, personal, and often offensive websites that go into an AI’s training data.” — 2023-04-21
- “In a few months, maybe a year, the first wave of AI-driven layoffs slash firings are going to hit the economy.” —Umair Haque … “Thanks to accounting conventions and tax laws dating back centuries, a robot doesn’t need to be better – or more efficient – than a human being at a task to make a business more profitable. It just needs to be 34% as good, or 11% as good, depending on that business’s accounting and amortization policies.” —John Carolus Sharp —2023-05-10