This is the fantasy of Stable Diffusion image junkies, ChatGPT tinkerers, and everyone who can’t get enough of their new AI content-generating toys. tool: You can get rich just by playing with AI.
“It’s happening, folks. ‘Ready Engineer’ is now a job title, earning between 175,000 and 300,000 a year,” TikTok user @startingname asserts in a gentle but firm tone.(opens in new tab) In his March TikTok post. To be eligible for the job, he says, you just have to “spend your time on algorithms.” In other words, it’s a six-figure job for anyone who likes messing around with generative AI.
of new york times called(opens in new tab) Prompt engineering is “a skill that long-time ChatGPT users can add to their resume.”of washington post called(opens in new tab) It’s “the hottest new job in tech”. Sinem Buber, chief economist at ZipRecruiter, told CNN.(opens in new tab) Thanks to rapid engineering, “more jobs will be created thanks to ChatGPT.”
Is this just hype? not exactly. There are indeed companies looking for ready-to-work engineers and offering generous salary packages. The most notable of these is Anthropic, home of the large language model Claude. The company is notorious for hiring an engineer quickly and offering his eye-popping six-figure sum.(opens in new tab) @startingname is mentioned in a TikTok video. But if you’re suffering from immediate engineer fever, take a sip of water, take a deep breath, and know that most immediate engineer jobs aren’t all that lucrative, and lucrative ones may not exist for very long. please.
Before shredding your medical school application, prioritizing your dream job of making funny pictures of the Pope in mid-journey(opens in new tab)After digging deeper into the work of a prompt engineer, here’s what we found:
What is a prompt engineer and why is this job so attractive?
Victor Sanh, a researcher at AI company Huggingface, who does AI-related research and builds applications, was one of the first academic researchers to approach prompt engineering and elaborated on it for Mashable. Did.
“Systems that have undergone reinforcement learning from human feedback (RLHF), such as ChatGPT, GPT-4, and Claude, tend to be more resilient to prompt formulations. [and] This basically means understanding that these systems were instructed in plain language. That’s the point. But Mr Sun continued. “They often make mistakes and either don’t understand the query correctly or realize that the query isn’t understood correctly.” The poorly specified query fails to provide a good response. ” That’s where prompt engineering comes in.
Users who find ChatGPT bugs can now earn up to $20,000
According to Sanh, another technique exists: chain of thought prompts, you need to split the query into subqueries. Sun called this “rapid engineering on steroids.”
For those who are fascinated by the idea of being an agile engineer, this may sound tediously technical. And there is no more fascinating topic in today’s tech space than generative AI. For about five years now, it’s felt like all the oxygen in tech news has been sucked out by a big shiny object called a cipher that is baffling and boring to the average person.But since OpenAI released his ChatGPT late last year, there’s been a whole new shiny object in the tech world that’s much easier for people, including kids.(opens in new tab) — To understand and get excited. If you think of ChatGPT as an app, it’s probably the most viral app of all time(opens in new tab).
So for one of the hundreds of millions of people who love playing with generative AI, the idea of becoming an agile engineer is as a job Perhaps being a video game tester sounds a bit like a job. Before you know video game testers can be underpaid and exhausting hard work.(opens in new tab)That is.
Unfortunately, if you get one of these jobs, i.e., one of the few people who are lucky enough to get a full-time job in exchange for real money and have “prompt engineer” on their resume. but you’re probably spending 8 hours a day spot researching. In Sanh’s words, if a system “doesn’t understand the query correctly or recognizes that the query is poorly specified,” a particular system needs to devise a way to generate a “appropriate response.” there is. It may sound monotonous, it may sound challenging, but that’s the crux of what you’re about to do.
Someone’s life may depend on you doing a good job as a quick engineer.
Benjamin Rader, a graduate research fellow in the Innovation and Digital Health Accelerator at Boston Children’s Hospital, is part of the team hiring immediate engineers. According to Rader, generative AI “allows doctors to create notes faster.” Boston Children’s Hospital needs “someone to help refine the demands of generative AI,” he explained. As such, the information we exhale is limited to hallucinations and specific to each particular task.
If you’re imagining that this will make ChatGPT write good knock-knock jokes, think again. “Each of these tasks will be really niche,” Rader said. The company’s internal systems are not trained on Wikipedia or Reddit, but instead on confidential memos regarding diagnoses, patient billing information, and comments between medical professionals regarding specific patients in very specific circumstances. increase. “We are a facility that often provides very specialized care,” he said.
Some Immediate Engineer jobs seem much less specialized, but also less profitable
Note that in some places the definition of the term “prompt engineer” is slightly looser than Sanh’s.1 current ad(opens in new tab) For example, Upwork is basically looking for someone to write a ton of cannabis articles.(opens in new tab). Content In his marketing and his SEO world, the job of people polishing up a lot of words in his day is nothing (the writer of the article you’re reading got his start in such a job). . GPT-3 and Jasper creating “his 800-1,500 word blog posts on various topics related to hemp, cannabis and CBD” are clearly a high-tech version of them. Needless to say, this is not one of his legendary six-figure jobs in the AI field.
What is a Good Immediate Engineer Qualification?
This career field is too new to apply any kind of credentialing process that hinders qualified applicants. In fact, Anthropic’s ad sounds like they’re considering Rand, who could justify themselves. “If you haven’t done much with prompt engineering yet, I would recommend spending some time experimenting with Claude or GPT3 and showing that you were able to get complex behavior from a well-crafted set of functions, and that’s a great way to get started with prompt engineering. You can best prove your skills, prompts,” the ad says. And hey, Who As long as you have access to one of the models, you can do that.
However, Matt Bell, a member of Anthropic’s technical staff, suggested to Mashable that a background in coding and machine learning (the process of “training” a model) would be of great help to applicants. “Having knowledge of coding helps when creating a systematic evaluation of prompts and helps you create prompts for coding work,” Bell said in an email.In addition, “I have ML [machine learning] Background helps you understand the overall strengths and weaknesses of these models. ”
Still, Bell hastened to add that the actual act of prompt engineering doesn’t require entering code, and Anthropic’s “best prompters are philosophers.”
In fact, “you need logic and reasoning to understand how the AI will react,” says Benjamin Rader of Boston Children’s Hospital. Applicants, polish Aristotle.
Will this kind of work continue into the future?
Huggingface researcher Sanh believes that the role of prompt engineers is temporary, and that vulnerabilities in AI systems are a well-known problem, and the companies creating these models and The organization says it is working to mitigate it. After all, you’re not supposed to type complex strings of stock phrases to interact with your chatbot. You just have to have a conversation.
Sanh likens agile engineers working directly with models to “a PR department preparing someone for a media interview.” Their role is like the unsophisticated candidates elected to make sure the system “just works”. He cited Microsoft’s Bing as an example. Bing’s chatbot famously melted down(opens in new tab) To new york times Written by columnist Kevin Roos, it was later published in a more palatable format and to a wider audience, with the exception of Meltdown. According to Sanh, that transformation probably involved a grand feat of rapid engineering. Because “it would likely have required a lot of interaction with the system to improve.”
Instead of chatbots, Rader’s team at Boston Children’s Hospital is training automated information processing systems for hospitals. “It will be important for a very long time to have trained people interacting with AI in the health system,” he said. It applies,” he said.
Either way, there are now lucrative new AI jobs. Some jobs may not exist in a few years, but many jobs may not exist in a few years thanks to AI. Please get in while you are feeling well.
