AI race for chatbot training data could run out of human-written text

Category: Technology/Innovations

Listening

Unlocking Word Meanings

Read the following words/expressions found in today’s article.

  1. exhaust / ɪgˈzɔst / (v.) – to use something completely so that nothing is left
    Example:

    If Shiela keeps spending too much, she will exhaust her savings before the end of the year.


  2. tap into / tæp ˈɪn tu / (phrasal v.) – to use something, such as money, knowledge, or other resources, in a way that helps achieve good results
    Example:

    The government is hoping to tap into renewable energy sources, such as wind and solar power, to reduce the use of fossil fuels.


  3. trajectory / trəˈdʒɛk tə ri / (n.) – a path or direction that something, such as a career or situation, follows as it develops, leading toward a specific result
    Example:

    The country’s economic trajectory has been positive, with consistent growth over the past decade.


  4. specialized / ˈspɛʃ əˌlaɪzd / (adj.) – developed or used for a particular purpose, function, or field of knowledge
    Example:

    This computer program is specialized for editing photos and creating digital art.


  5. degraded / dɪˈgreɪ dɪd / (adj.) – reduced in quality, value, or strength
    Example:

    The teacher expressed concern over the degraded performance of students in the latest assessments.


Article

Read the text below.

Artificial intelligence (AI) systems like ChatGPT could soon run out of what keeps making them smarter—the tens of trillions of words people have written and shared online.


A new study released by research group Epoch AI projects that tech companies will exhaust the supply of publicly available training data for AI language models by roughly the turn of the decade—sometime between 2026 and 2032.


In the short term, tech companies like ChatGPT-maker OpenAI and Google are racing to secure and sometimes pay for high-quality data sources to train their AI large language models—for instance, by signing deals to tap into the steady flow of sentences coming out of Reddit forums and news media outlets.


In the longer term, there won’t be enough new blogs, news articles, and social media commentary to sustain the current trajectory of AI development, putting pressure on companies to tap into sensitive data now considered private—such as emails or text messages—or rely on less-reliable “synthetic data” spit out by the chatbots themselves.


Tamay Besiroglu, an author of the study, said AI researchers realized more than a decade ago that aggressively expanding two key ingredients—computing power and vast stores of internet data—could significantly improve the performance of AI systems.


“I think it’s important to keep in mind that we don’t necessarily need to train larger and larger models,” said Nicolas Papernot, an assistant professor of computer engineering at the University of Toronto and researcher at the nonprofit Vector Institute for Artificial Intelligence.


Papernot, who was not involved in the Epoch study, said building more skilled AI systems can also come from training models that are more specialized for specific tasks. But he has concerns about training generative AI systems on the same outputs they’re producing, leading to degraded performance known as “model collapse.”


Training on AI-generated data is “like what happens when you photocopy a piece of paper and then you photocopy the photocopy. You lose some of the information,” Papernot said. Not only that, but Papernot’s research has also found it can further encode the mistakes, bias, and unfairness that’s already baked into the information ecosystem.


This article was provided by The Associated Press.


Viewpoint Discussion

Enjoy a discussion with your tutor.

Discussion A

  • Tens of trillions of words people have written and shared online that are used for AI training may soon run out. Given that possibility, what is the future of AI? Do you think it will survive? Why or why not? Discuss.
  • Given the possibility that training data may run out and may affect the effectiveness of AI, how much should we rely on AI? Discuss.

Discussion B

  • Companies are pressured to tap into sensitive data now considered private, such as emails or text messages, to train AI. Do you think this is ethical? What do you think are the consequences of this action (ex. there will be privacy concerns, there will be misuse of data)? Discuss.
  • How might public resistance to the use of private data for AI training impact the development and adoption of AI technologies (ex. it will delay the development of AI technologies)? Discuss.