This story originally appeared in The Algorithm, our weekly AI newsletter. To get stories like this delivered to your inbox first,sign up here.
I've been thinking a lot this week about the human work behind smart AI models.
The secret to making AI chatbots sound smart and spew less toxic nonsense is to use a technique calledreinforcement learning from human feedback, which uses people's input to improve the model's responses.
It relies on a small army of human data annotators to assess whether a string of text makes sense and sounds smooth and natural. They decide whether a response should be kept in the AI model's database or deleted.
Even the most impressive AI chatbots require thousands of hours of human work to behave as their creators want them to, and even then they do so in unreliable ways.The work can be brutal and disruptive, as we'll hear this week when the ACM Conference on Fairness, Accountability and Transparency (FAccT) kicks off. It's a conference that brings together research on topics I like to write about, such as how to make AI systems more accountable and ethical.
InpanelI'm looking forward to hanging out with AI ethics pioneer Timnit Gebru, who used to head Google's AI ethics department before he was fired. Gebru will talk about how data workers in Ethiopia, Eritrea and Kenya are being exploited to clean up hate and misinformation online. Data recipients in Kenya were e.gpay less than $2 an hourto filter lots of disturbing content about violence and sexual abuse to make ChatGPT less toxic. These workers are nowUnionto get better working conditions.
In a series in MIT Technology Review last year, we explored how AI creates onecolonial new world order, ycomputer workers bear the burden. Elucidating exploitative work practices around AI has become even more urgent and important with the emergence of popular AI chatbots such as ChatGPT, Bing and Bard and image-generating AI such as DALL-E 2 and Stable Diffusion.
Data annotators are involved in all stages of AI development, from training models to verifying their results to providing feedback that allows a model to be fine-tuned after it is released.They are often forced to work at an incredibly fast pace to meet high targets and tight deadlines, says Srravya Chandhiramowuli, a PhD researcher who studies employment practices in data work at City, University of London.
"This notion that you can build these systems at scale without human intervention is an absolute fallacy," says Chandhiramowuli.
Data annotators give AI models the important context they need to make decisions at scale and appear sophisticated.
Chandhiramowuli tells me of a case where a computer operator in India had to distinguish between images of soda bottles and pick out the ones that resembled Dr. Pepper. But Dr. Pepper is not a product sold in India and it was the data recorder's responsibility to find out.
The expectation is that the annotators will discover the values that are important to the company, says Chandhiramowuli. "Not only are they learning these distant things that don't make sense to them, but they're also figuring out not only what the other contexts are, but what the priorities of the system they're building are," she says.
In fact, we are all data workers for big tech companies, whether we realize it or not.argue researchers from the University of California, Berkeley, the University of California, Davis, the University of Minnesota and Northwestern University in anew paperpresented at FAccT.
The text and image AI models are trained using huge datasets that have been retrieved from the internet. This includes oursPersonal Informationycopyrighted works of artists, and the data that we created is now forever part of an AI model built to make a business make money. We unwittingly contribute our work for free by uploading our images to public websites, voting on comments on Reddit, tagging images in reCAPTCHA, or searching online.
Currently, the balance of power is heavily skewed in favor of some of the world's largest tech companies.
To change that, we need nothing less than a data revolution and regulation.The researchers argue that one way people can take back control of their online lives is to advocate for transparency about how data is used and find ways to give people the right to provide feedback and share revenue from its use. data.
Although this data work forms the backbone of modern AI, data work remains chronically undervalued and invisible around the world, and wages remain low for annotators.
"There is absolutely no recognition of what the contribution of data work is," says Chandhiramowuli.
The future of generative artificial intelligence and business
What are you doing on Wednesday? Why don't you join me and MIT Technology Review Senior AI Editor Will Douglas HeavenFollow, where a large panel of experts will join us to analyze how the AI revolution will change business.
My sessions will look at AI in cyber security, the importance of data and the new rules we need for AI.Tickets are still available here.
to whet the appetite, my colleague David Rotman has a deep dive into generative artificial intelligence and how it's going to change the economy.read it here.
Even deeper learning
DeepMind's game AI just found another way to make code faster
Using a new version of AlphaZero AI for games called AlphaDev, the UK-based company (recently renamed Google DeepMind after a merger with its sister company's AI lab in April) has discovered a way to classify items in a list of up to 70%. faster than the best existing method. He also found a way to speed up a key algorithm used in cryptography by 30%.
Why this matters:As the computer chips that power AI models approach their physical limits, computer scientists need to find new and innovative ways to optimize data processing. These algorithms are among the most common building blocks of software. Small speedups can make a big difference, reducing costs and saving energy.Read more from Will Douglas Heaven here.
Bits y bytes
Ron DeSantis' ad uses AI-generated images of Donald Trump and Anthony Fauci
The US presidential election is getting complicated. Exhibit A: A campaign supporting Ron DeSantis as the 2024 Republican presidential nominee has used an AI-generated deepfake to attack his rival Donald Trump. The photo shows Trump kissing Anthony Fauci, a former White House chief medical adviser who is hated by many on the right.AFP)
Humans are biased, but generative AI is worse
This visual study shows how the open source text-to-image stable diffusion model reinforces stereotypes about race and gender. The piece is a great display ofinvestigationshows that the AI model presents a more biased view of the world than reality. For example, women made up only 3% of the images generated for the keyword “judge”, while 34% of US judges are actually women. (Bloomberg)
Meta is rolling out generative AI for everything
After a rocky year of layoffs, Meta CEO Mark Zuckerberg told staff that the company intends to integrate generative AI into its flagship products like Facebook and Instagram. For example, people will be able to use text prompts to edit photos and share them on Instagram Stories. The company is also developing AI assistants or trainers that people can interact with. (Los New York Times)
A successful use of generative AI
see someonefix things using generative AIin image editing software.
Although the implications of the development of AI-driven software are far-reaching, it is still important for knowledge workers to remain optimistic about their future. Many highly skilled jobs still require creative problem-solving, and AI will never replace the human touch.Will AI replace the needs for human to work? ›
While AI can replace some tasks, it cannot replace human problem-solving skills. Therefore, combining the strengths of AI and human curiosity is necessary to achieve outstanding results in scientific pursuits.How can AI and humans work together? ›
AI Can Help Human Focus
AI has the potential to eliminate tedious tasks and increase attention to crucial ones. People can become specialists in their roles in particular fields if they spend more time doing the things that matter.
AI supports individual performance within workforce ecosystems. Businesses are increasingly relying on augmented reality/virtual reality (AR/VR) technologies, for instance, to enhance individual and team performance. These technologies promise to improve worker safety in some workplace environments.What jobs Cannot be replaced by AI? ›
As such, jobs that require high emotional intelligence, such as therapists, social workers, and nurses, are not likely to be replaced by AI. Specialized Professionals: Jobs that require deep expertise in a particular field, such as doctors, lawyers, and scientists, are less likely to be fully replaced by AI.Is AI going to take over my job? ›
"Additionally, the development and maintenance of AI systems require human expertise and oversight, meaning that AI will not completely replace the need for human workers in these fields," the bot wrote. AI expert Kai Fu Lee warned that 40 percent of the world's jobs could be completed by machines in the next 15 years.What jobs will AI replace first? ›
- Jobs most impacted by AI. Advertisement. ...
- Coders/programmers. ...
- Writers. ...
- Finance professionals. ...
- Legal workers. ...
- Researchers. ...
- Customer service. ...
- Data entry and analysis.
- Data Scientist. Data scientists are a new generation of data analytics experts, already among the most in-demand professions for AI. ...
- Artificial Intelligence And Machine Learning Engineers. ...
- Data Labeling Professionals. ...
- Hardware Specialists For Artificial Intelligence.
AI lacks many of the essential human traits that are required in various fields such as creativity, emotional intelligence, contextual understanding, common sense, adaptability, ethics, intuition, physical dexterity, interpersonal skills, adaptability to change, imagination, and free will.What are two things AI can do that humans cant? ›
AI can filter email spam, categorize and classify documents based on tags or keywords, launch or defend against missile attacks, and assist in complex medical procedures. However, if people feel that AI is unpredictable and unreliable, collaboration with this technology can be undermined by an inherent distrust of it.
AI rapidly analyzes, categorizes, and classifies millions of data points, and gets smarter with each iteration. Learning through feedback from the accumulation of data is different from traditional human learning, which is generally more organic. After all, AI can mimic human behavior but cannot create it.Can AI do everything a human can do? ›
While AI is designed to replace manual labor with a more effective and quicker way of doing work, it cannot override the need for human input in the workspace. In this article, you will see why humans are still immensely valuable in the workplace and cannot be fully replaced by AI.Is artificial intelligence a threat to humans? ›
Abstract. While artificial intelligence (AI) offers promising solutions in healthcare, it also poses a number of threats to human health and well-being via social, political, economic and security-related determinants of health.What jobs are most at risk of being replaced by AI in the future? ›
- Entry-Level admin roles. Administrative roles are perhaps most at risk of being replaced by AI tools currently available, such as ChatGPT. ...
- Data entry clerks. ...
- Software Engineers and Coders. ...
- Customer Service Representatives. ...
- Legal assistants. ...
- Copywriting and content production. ...
- Graphic designers. ...
Long answer short —AI models like Chat GPT can be a valuable tool for data scientists, but they cannot replace the important role that data scientists play in various industries. This is true for most of the roles.What are the 5 things that robots can do that humans can t? ›
- Handling tedium. ...
- Extreme sensing. ...
- Strength and speed. ...
- Unwavering focus. ...
- Perfect, objective recall.
Since ChatGPT's release in November of 2022, speculation has grown over whether or not the role of a data analyst could eventually be replaced by generative AI (ChatGPT, Bard, and Bing Chat are among the large language models included in this classification).
Speaking via video link to a summit in London, Musk said he expects governments around the world to use AI to develop weapons before anything else. Elon Musk has hit out at artificial intelligence (AI), saying it is not "necessary for anything we're doing".Why are people worried about AI? ›
AI lacks the ability to consider these intangible factors and may make decisions solely based on pre-programmed algorithms or data inputs, which could lead to unintended consequences or even catastrophic errors. Secondly, the use of AI in military decision-making raises significant ethical concerns.Will AI destroy more jobs than it creates? ›
Like many revolutionary technologies before it, AI is likely to eliminate jobs. But, as has been the case in the past, experts argue, AI will likely offset much of that by spurring the creation of new jobs in addition to enhancing many existing jobs. The big question is: what sort of jobs?
"If I were to pick out five areas where I think AI will have the most significant impact on the future of work, I'd say it would be: manufacturing, customer service, retail, finance and healthcare. "Manufacturing is one of the industries already seeing the biggest impact from AI.What industries are not affected by AI? ›
According to the report, jobs in agriculture, mining and manufacturing are the least exposed to generative AI, while jobs in the information processing industries, like IT, are the most exposed because jobs that use "programming and writing skills" are more closely related to GPT's capabilities.What is the next technology after AI? ›
Experts and futurists predict that the next trend after AI could be quantum computing, augmented reality, blockchain technology, the Internet of Things (IoT), or a combination of these technologies.What jobs will be around forever? ›
- Lawyer. As much as legal analysis and case preparation will become increasingly automated, we are a long way away from robots representing or cross-examining us in the courtroom! ...
- HR roles. ...
- Tradespeople. ...
- IT systems analysts. ...
- Medical professionals.
- Travel agents. Technology has undermined the role of the travel agent. ...
- Cashiers. With more than 3.3 million people working in this capacity, the job of cashier isn't going to disappear anytime soon. ...
- Bank tellers. ...
- Drivers. ...
- Newspapers. ...
- Fast-food workers. ...
- Telemarketing. ...
- Warehouse workers.
AI Will NOT Replace Independent Insurance Agents
The short answer is that artificial intelligence is highly unlikely to replace independent insurance agencies. Some things require a human touch, and insurance is one of those. So, your career is safe.
Sometimes, the AI system may be solely responsible. In other cases, the humans who created or are using the AI system may be partially or fully responsible. Determining who's responsible for an AI mistake can be difficult, and it may require legal experts to determine liability on a case-by-case basis.Can AI exist without humans supporting it? ›
They depend on the codes humans write, so they can't think critically through complex scenarios. Further, AI processes data that humans enter into the system. Without human interventions, machines will lie idle.What do humans do better than AI? ›
They are more adept at comprehending the subtleties of a situation from different perspectives which is crucial for guaranteeing quality, which AI is not able to do. Further, humans have emotional intelligence which helps them recognise and cater to the needs and feelings of others.Will AI replace white-collar workers? ›
“A.I. will increasingly replace repetitive jobs, not just for blue-collar work, but a lot of white-collar work,” Kai-Fu Lee, a venture capitalist focused on A.I., said in 2019.
In conclusion, it is projected that the growth of AI and automation would alter the relevance of blue-collar versus white-collar jobs. Due to the skills they require and the sense of purpose they provide, blue-collar professions will become more desirable, whereas many white-collar jobs face the threat of automation.Will AI be able to replace teachers? ›
I recently asked Bard, Google's conversational chatbot, whether artificial intelligence would replace teachers. Here's what it said, “It is unlikely that AI will completely replace teachers in the near future.”How does AI affect knowledge workers? ›
AI Will Bring Higher Productivity and Profits
“That is a leadership choice, and it's also execution.” The transformation will pile pressure on the labor force, especially for higher-wage knowledge workers whose activities “were previously considered to be relatively immune from automation,” the report said.
Therefore, while AI can be a helpful tool in therapy, it is unlikely to replace the role of human therapists, social workers, or even massage therapists anytime soon.Will AI replace healthcare workers? ›
Already, AI is being used to streamline administrative tasks, answer patient questions, and for machine learning; its likely that in the near future there will be more AI scribes and virtual nursing assistants. As the technology continues developing, AI will be used to supplement care but it can't replace doctors.Will AI replace medical professionals? ›
Will artificial intelligence replace my doctor? A. Not in my lifetime, fortunately! And the good news is that artificial intelligence (AI) has the potential to improve your doctor's decisions, and to thereby improve your health — if we are careful about how it is developed and used.Can lawyers be replaced by AI? ›
Professor Eric Talley of Columbia Law School, who recently taught a course on Machine Learning and the Law, says AI won't replace lawyers but will instead complement their skills, ultimately saving them time, money and making them more effective.Will AI replace nurses? ›
In conclusion, while the development and advancement in technology and AI is a boon and helps reduce (human)physician/nursing errors in diagnosis and treatment of patients , I believe that it cannot replace humans completely and should have a collaborative approach and be used as a tools/means to aid physicians/nurses ...Will AI make education obsolete? ›
AI tools can certainly enhance and support learning experiences. However, they cannot replace the critical role that teachers play in fostering meaningful relationships with students, providing mentorship and guidance, and creating supportive and engaging learning environments.How AI will impact the future of work and workers? ›
Increased Efficiency: AI can automate many routine tasks and workflows, freeing up workers to focus on higher-level tasks and increasing productivity. Improved Accuracy: AI systems can process large amounts of data quickly and accurately, reducing the risk of errors.
AI has the ability to improve safety in the workplace by reducing human error, automating tasks, and providing employees with real-time feedback. When it comes to safety in the workplace, AI can be used in a number of ways. For example, AI can be used to monitor employee behavior and identify potential risks.What is the impact of AI on job loss? ›
In March, investment bank Goldman Sachs predicted in a report that AI could eventually replace 300 million full-time jobs globally and affect nearly one-fifth of employment — with a particular hit to white-collar jobs often considered automation-proof, such as administrative and legal professions.