6 hrs ago
IIT Professor Challenges Calls for Global AI Slowdown
Professor B Ravindran says today’s AI systems are computer programs, not conscious beings planning to hurt people.
They follow goals and instructions provided by humans.
For an AI system to cause a catastrophe, many people would first have to give it dangerous access or fail to control it.
He says this means humans still have many chances to stop a bad outcome.
Ravindran thinks companies should improve safety controls, testing environments and cybersecurity.
Governments can create liability rules, but companies understand their systems best.
He believes large, general-purpose AI models are harder and more expensive to make safe.
He supports companies slowing down their own risky experiments, but not asking the entire world to stop AI research.
He also says some calls for a slowdown may be connected to competition, investment and data-centre costs.
Professor B Ravindran says AI is not independently plotting to eliminate humanity.
He argues catastrophic outcomes would require multiple human decisions and failures along the way.
Ravindran says companies, not governments, currently have the technical capacity to build AI safeguards.
He believes frontier AI companies may favor a slowdown partly to protect their lead and reduce escalating costs.
He supports stronger testing, cybersecurity controls and kill switches, but rejects a worldwide halt to AI research.
- Who
- Professor B Ravindran, Geoffrey Hinton, Dario Amodei and other AI industry leaders are central to the discussion.
- What
- The interview examines whether AI development should slow down and how serious the risks of advanced AI are.
- Where
- The interview concerns the global AI industry and was conducted with a professor at IIT Madras.
- When
- The discussion refers to events earlier this year and developments described as occurring this month.
- Why
- Recent warnings about AI potentially becoming uncontrollable, along with rising development and safety costs, have prompted calls for a slowdown.
Slowdown Advocates
Targeted Safety Advocates
Whether AI development should slow
Slowdown Advocates
Figures including Dario Amodei, Geoffrey Hinton, Elon Musk and Sam Altman have supported slowing or pausing aspects of AI development because advanced systems could become dangerous.
Targeted Safety Advocates
B Ravindran says a global slowdown is not warranted and that companies calling for one can slow their own work while improving safety.
Nature of the threat
Slowdown Advocates
AI could eventually become extremely capable and potentially go out of control, creating catastrophic risks.
Targeted Safety Advocates
Ravindran says current AI is not sentient and does not independently plan to kill humans; catastrophic outcomes would require many human decisions and inadequate protections.
Reason for industry warnings
Slowdown Advocates
Warnings may reflect genuine concerns that increasingly capable models are becoming harder to control.
Targeted Safety Advocates
Ravindran says companies may also be protecting their competitive lead, avoiding the rising cost of safeguards or facing limits involving data centres and computing resources.
Key facts
- Interview subject
- Professor B Ravindran of IIT Madras
- Core argument
- AI does not have an established innate drive or conscience to harm humanity.
- Main risk
- Human decisions, inadequate safeguards and unsafe access could enable harmful outcomes.
- Safety responsibility
- Companies currently have the strongest technical ability to implement AI guardrails.
- Government role
- Governments can establish legal guardrails, liability rules and incentives for safety work.
- Preferred safeguards
- Stronger sandboxing, cybersecurity controls and kill switches for AI models.
- Development concern
- More capable, general-purpose models are increasingly expensive and difficult to safeguard.
Quotes
Professor B Ravindran
IIT Madras professor and head of its data science and responsible AI centers
“The legal guardrails, clear liability directions, etc can be given by governments, but the companies have to work on the technology to impose the guardrails.”
rediff.com
“I think it is blind fear mongering when you say there is a 10 per cent extermination rate within the next five years or so.”
rediff.com
Sources
Related news
AI Leaders Debate Slowing Development Amid Fears of Human Extinction
AI Safety Debate Intensifies as Leaders Weigh Development Pace
AI Kill Switch Debate Pits Safety Against Industry Self-Regulation
Altman and Musk Back Amodei’s Call to Slow AI
AI Safety Debate Grows Over Calls For Development Slowdown
Anthropic CEO Dario Amodei Urges AI Industry to Slow Down
Anthropic CEO Warns UN of AI Risks to Humanity
Prince Harry Warns AI Seeks Children’s Attachment Beyond Attention
Anthropic Says Claude Found CRISPR-Like System in Bacteriophage DNA
AI Leaders Urge UN to Regulate Existential Risks
