Some researchers think that AI could eventually achieve general intelligence, matching and even exceeding humans on most tasks.Credit: Charles Taylor/Alamy
Will an artificial intelligence (AI) superintelligence appear suddenly, or will scientists see it coming, and have a chance to warn the world? Thats a question that has received a lot of attention recently, with the rise of large language models, such as ChatGPT, which have achieved vast new abilities as their size has grown. Some findings point to emergence, a phenomenon in which AI models gain intelligence in a sharp and unpredictable way. But a recent study calls these cases mirages artefacts arising from how the systems are tested and suggests that innovative abilities instead build more gradually.
I think they did a good job of saying nothing magical has happened, says Deborah Raji, a computer scientist at the Mozilla Foundation who studies the auditing of artificial intelligence. Its a really good, solid, measurement-based critique.
The work was presented last week at the NeurIPS machine-learning conference in New Orleans.
Large language models are typically trained using huge amounts of text, or other information, whch they use to generate realistic answers by predicting what comes next. Even without explicit training, they manage to translate language, solve mathematical problems and write poetry or computer code. The bigger the model is some have more than a hundred billion tunable parameters the better it performs. Some researchers suspect that these tools will eventually achieve artificial general intelligence (AGI), matching and even exceeding humans on most tasks.
ChatGPT broke the Turing test the race is on for new ways to assess AI
The new research tested claims of emergence in several ways. In one approach, the scientists compared the abilities of four sizes of OpenAIs GPT-3 model to add up four-digit numbers. Looking at absolute accuracy, performance differed between the third and fourth size of model from nearly 0% to nearly 100%. But this trend is less extreme if the number of correctly predicted digits in the answer is considered instead. The researchers also found that they could also dampen the curve by giving the models many more test questions in this case the smaller models answer correctly some of the time.
Next, the researchers looked at the performance of Googles LaMDA language model on several tasks. The ones for which it showed a sudden jump in apparent intelligence, such as detecting irony or translating proverbs, were often multiple-choice tasks, with answers scored discretely as right or wrong. When, instead, the researchers examined the probabilities that the models placed on each answer a continuous metric signs of emergence disappeared.
Finally, the researchers turned to computer vision, a field in which there are fewer claims of emergence. They trained models to compress and then reconstruct images. By merely setting a strict threshold for correctness, they could induce apparent emergence. They were creative in the way that they designed their investigation, says Yejin Choi, a computer scientist at the University of Washington in Seattle who studies AI and common sense.
Study co-author Sanmi Koyejo, a computer scientist at Stanford University in Palo Alto, California, says that it wasnt unreasonable for people to accept the idea of emergence, given that some systems exhibit abrupt phase changes. He also notes that the study cant completely rule it out in large language models let alone in future systems but adds that "scientific study to date strongly suggests most aspects of language models are indeed predictable.
Raji is happy to see the community pay more attention to benchmarking, rather than to developing neural-network architectures. Shed like researchers to go even further and ask how well the tasks relate to real-world deployment. For example, does acing the LSAT exam for aspiring lawyers, as GPT-4 has done, mean that a model can act as a paralegal?
The work also has implications for AI safety and policy. The AGI crowd has been leveraging the emerging-capabilities claim, Raji says. Unwarranted fear could lead to stifling regulations or divert attention from more pressing risks. The models are making improvements, and those improvements are useful, she says. But theyre not approaching consciousness yet.
Originally posted here:
Will superintelligent AI sneak up on us? New study offers reassurance - Nature.com
- How would a Victorian author write about generative AI? - Verdict [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Is the current regulatory system equipped to deal with AI? - The Hindu [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Tim Sweeney, CD Projekt, and Other Experts React to AI's Rise, and ... - IGN [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- GCHQ chiefs warning to ministers over risks of AI - The Independent [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Solving The Mystery Of How ChatGPT And Generative AI Can Surprisingly Pick Up Foreign Languages, Says AI Ethics And AI Law - Forbes [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- SenseAuto Empowers Nearly 30 Mass-produced Models Exhibited at Auto Shanghai 2023 and Unveils Six Intelligent Cabin Products - Yahoo Finance [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- AI Dangers Viewed Through the Perspective of Don't Look Up - BeInCrypto [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- ChatGPT, artificial intelligence, and the news - Columbia Journalism Review [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Unpacking AI: "an exponential disruption" with Kate Crawford: podcast and transcript - MSNBC [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Is 'Generative' AI the Way of the Future? - Northeastern University [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- GPT-4 Passes the Bar Exam: What That Means for Artificial ... - Stanford Law School [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Is artificial intelligence approaching science fiction? - The Trail - The Puget Sound Trail [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- Towards Artificial General Intelligence, ChatGPT 5 is on Track - Analytics Insight [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- New AGI hardware in progress for artificial general intelligence - Information Age [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- ChatGPT is impressive, but it may slow the emergence of AGI - TechTalks [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- How smart is ChatGPT really and how do we judge intelligence in AIs? - New Scientist [Last Updated On: August 18th, 2024] [Originally Added On: April 20th, 2023]
- AI Singapore and the Digital and Intelligence Service Sign ... - MINDEF Singapore [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- Artificial Intelligence Will Take Away Jobs and Disrupt Society, says Zerodha CEO Nithin Kamath - DATAQUEST [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- China's State-Sponsored AI Claims it Will Surpass ChatGPT by End ... - Tom's Hardware [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- AI At The Crossroads: Navigating Job Displacement, Ethical Concerns, And The Future Of Work - Forbes [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- I created a billion-pound start-up business Elon Musk & Jeff Bezos asked to meet me heres the secret to... - The Sun [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- Opinion | We Need a Manhattan Project for AI Safety - POLITICO [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- As AutoGPT released, should we be worried about AI? - Cosmos [Last Updated On: August 18th, 2024] [Originally Added On: May 14th, 2023]
- The Role of Artificial Intelligence in the Future of Media - Fagen wasanni [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- The Threat Of Climate Misinformation Propagated by Generative AI ... - Unite.AI [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- AI and the Next Phase of Human Evolution: What Can We Expect? - Fagen wasanni [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- IT pros mull observability tools, devx and generative AI - TechTarget [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- The future of learning and skilling with AI in the picture - Chief Learning Officer [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- The Economic Case for Generative AI and Foundation Models - Andreessen Horowitz [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Exploring the future of AI: The power of decentralization - Cointelegraph [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Convergence of Brain-Inspired AI and AGI: Exploring the Path to ... - Newswise [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Past, Present, Future: AI, Geopolitics, and the Global Economy - Tech Policy Press [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Development of GPT-5: The Next Step in AI Technology - Fagen wasanni [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Executive Q&A: Andrew Cardno, QCI - Indian Gaming [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- The Future is Now: Understanding and Harnessing Artificial ... - North Forty News [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- The Twin Convergence: AGI And Superconductors Ushering Humanity's Inflection Point - Medium [Last Updated On: August 4th, 2023] [Originally Added On: August 4th, 2023]
- Moonshot: Coexisting with AI holograms - The Edge Malaysia [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- A large computing cluster at sea might have big implications for AI ... - XDA Developers [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- ChatGPT or not ChatGPT? That was the question, briefly, as ... - GeekWire [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- The Best ChatGPT Prompts Are Highly Emotional, Study Confirms - Tech.co [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- The impact of AI and Language Models - Girton College [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Startup gaining investment traction for AI clinician productivity tool - Mobihealth News [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- What OpenAI's latest batch of chips says about the future of AI - Quartz [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- How AI Ecosystems Are Transforming the Future of Business - Entrepreneur [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Understanding Artificial Intelligence: Definition, Applications, and ... - Medium [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Europe's weaknesses, opportunities facing the AI revolution - EURACTIV [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- How the AI Executive Order and OMB memo introduce ... - Brookings Institution [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Artificial intelligence: the world is waking up to the risks - InCyber [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- How to win the artificial general intelligence race and not end ... - The Strategist [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- AI 2023: risks, regulation & an 'existential threat to humanity' - RTE.ie [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Game-playing DeepMind AI can beat top humans at chess, Go and poker - New Scientist [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- Sam Altman Seems to Imply That OpenAI Is Building God - Futurism [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- OpenAI's six-member board will decide 'when we've attained AGI' - VentureBeat [Last Updated On: August 18th, 2024] [Originally Added On: November 16th, 2023]
- The Most Important AI Innovations of 2024 | by AI News | Dec, 2023 - DataDrivenInvestor [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- Forget Dystopian Scenarios AI Is Pervasive Today, and the Risks Are Often Hidden - The Good Men Project [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- The Era of AI: 2023's Landmark Year - CMSWire [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- The Impact of OpenAIs GPT 5. A New Era of AI | by Courtney Hamilton | Dec, 2023 - Medium [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- What Is Artificial Intelligence? From Software to Hardware, What You Need to Know - ExtremeTech [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- AI consciousness: scientists say we urgently need answers - Nature.com [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- AI Technologies Set to Revolutionize Multiple Industries in Near Future - Game Is Hard [Last Updated On: August 18th, 2024] [Originally Added On: December 27th, 2023]
- Amazon reportedly preparing paid Alexa version powered by its own Titan AI model - SiliconANGLE News [Last Updated On: August 18th, 2024] [Originally Added On: May 25th, 2024]
- OpenAI departures: Why cant former employees talk, but the new ChatGPT release can? - Vox.com [Last Updated On: August 18th, 2024] [Originally Added On: May 25th, 2024]
- Meta AI Head: ChatGPT Will Never Reach Human Intelligence - PYMNTS.com [Last Updated On: August 18th, 2024] [Originally Added On: May 25th, 2024]
- What is artificial general intelligence, and is it a useful concept? - New Scientist [Last Updated On: August 18th, 2024] [Originally Added On: May 25th, 2024]
- 22 jobs artificial general intelligence (AGI) may replace and 10 jobs it could create - Livescience.com [Last Updated On: August 18th, 2024] [Originally Added On: May 25th, 2024]
- Why AI Won't Take Over The World Anytime Soon - Bernard Marr [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- OpenAI announces new Safety and Security Committee as the AI race hots up and concerns grow around ethics - TechRadar [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- Responsible AI needs further collaboration - Chinadaily.com.cn - China Daily [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- OpenAI says it's charting a "path to AGI" with its next frontier AI model - ITPro [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- The AI revolution is coming to robots: how will it change them? - Nature.com [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- Can AI ever be smarter than humans? | Context - Context [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]
- OpenAI revives its robotic research team, plans to build dedicated AI - Interesting Engineering [Last Updated On: August 18th, 2024] [Originally Added On: June 3rd, 2024]