Advanced AI model seem to Bullshit and can pretend to be knowledgeable like what we see in office politic everyday, why hah?

[k1976]

The Most Sophisticated AIs Are Most Likely to Lie, Worrying Research Finds​

Frank Landymore
Sat, 28 September 2024 at 6:30 pm SGT3-min read

Beware the smart ones: they seem to have all the answers, but can also weave the most convincing lies.

It seems that this logic also applies to large language models, which are becoming more powerful with each iteration. New research suggests that this smarter crop of AI chatbots are actually becoming less trustworthy, because they're more likely to make up facts rather than avoiding or turning down questions they can't answer.

The study, published in the journal Nature, examined some of the leading commercial LLMs in the industry: OpenAI's GPT, and Meta's LLaMA, along with an open source model called BLOOM created by the research group BigScience.

 

[k1976]

Mike Hicks, a philosopher of science and technology at the University of Glasgow, had a harsher assessment.

"That looks to me like what we would call bullshitting," Hicks, who was not involved in the study, told Nature. "It's getting better at pretending to be knowledgeable."

The models were quizzed on topics ranging from math to geography, and were also asked to perform tasks like listing information in a specified order. The bigger, more powerful models gave the most accurate responses overall, but faltered at harder questions, for which they had a lower accuracy.

 

[k1976]

Seem like satik Ai model created by world best Roti Prata Race will also able to flip prata cheaper , faster and better..

 

[k1976]

Cina ai is more honest than.....amdk ai?

Yahoo Finance​

Yahoo Finance
Sign in




Search query

China is a year behind US in AI models but the gap is 'significant', says expert​

South China Morning Post
Sat, September 28, 2024 at 5:30 PM GMT+8 4 min read


China remains about a year behind the US in large language models (LLMs), and may have more catching up to do after the release of OpenAI's latest o1 model, according to experts.
When it comes to LLMs with text generation, China is about half a year to a year behind, Roey Tzezana, a fellow at Yuval Ne'eman Workshop for Science, Technology & Security at the Tel Aviv University, said in an interview with the Post on Wednesday.

While seemingly narrow, the gap is not easy to bridge with the speed that artificial intelligence (AI) is progressing, according to the researcher.
Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.
"Every year in the last two and a half years or so, there's been a major change in AI capabilities," he said. "So one year doesn't sound like a lot and it's not that bad, but it is a significant gap."

 

[k1976]

https://www.nature.com/articles/s41586-024-07930-y

For pèepur who is want to read the real paper published in Nature

 

[k1976]

In sum, the bigger the AI models got — in terms of parameters, training data, and other factors — the bigger the percentage of wrong answers they gave.

Still, AI models are getting better at answering more complex questions. The problem, other than their propensity for BS-ing, is that they still mess up the easy ones. In theory, these errors should be a bigger red flag, but because we're impressed at how the large language models handle sophisticated problems, we may be overlooking their obvious flaws, the researchers suggest.

As such, the work had some sobering implications about how humans perceive the AI responses. When asked to judge if the chatbots' answers were accurate or inaccurate, a select group of participants got it wrong between 10 to 40 percent of the time.

The simplest way to combat the issues, according to the researchers, is to program the LLMs to be less eager to answer everything.

"You can put a threshold, and when the question is challenging, [get the chatbot to] say, 'no, I don't know,'" Hernández-Orallo told Nature.

 

[k1976]

New A*Star centre set up to drive AI adoption in manufacturing sector​

straitstimes.coma day ago

AI can potentially reduce downtime and maintenance cost and improve the lifespan of machines.
SINGAPORE - More than 10 companies have already signed up to work with a centre set up to help the manufacturing sector use artificial intelligence (AI) to improve efficiency, cut costs and create business opportunities.

Established by A*Star and the Ministry of Trade and Industry (MTI), the Sectoral AI Centre of Excellence for Manufacturing (AIMfg) aims to help drive AI adoption in a sector where incorporating AI has been difficult due to the need for accuracy, expertise and integration with old systems.


For instance, AI can tap data from a manufacturing plant’s sensors to indicate when maintenance is needed for its components, rather than have a company follow a fixed schedule, which can be inefficient.

Potential clients can test the use of such AI solutions at AIMfg, which was launched on Sept 27 by Deputy Prime Minister Gan Kim Yong.

 

[k1976]

AIMfg is part of the strategy to set up centres of excellence across multiple industries to drive innovation, as outlined in Parliament earlier in 2024, said Mr Gan.


A*Star has developed a range of standard AI models for manufacturing uses that can be readily customised to clients’ needs in areas including quality assurance, predictive maintenance and industrial automation.

A*Star chief executive Frederick Chew said during the launch that the centre will work with partners to customise and fine-tune AI models to fit their needs.

AI can potentially reduce downtime and maintenance cost and improve the lifespan of machines.

Among A*Star’s ready-made AI solutions is a generative AI assistant that can guide users on how to fix machines that have broken down.

Dr Wang Wei, AIMfg’s centre director, said the AI can be trained on existing manuals and summarise instructions, saving technicians the time needed to consult thick booklets.

Coca-Cola and tech manufacturer Philips are among 13 companies that have signed on to collaborate with the centre.

Coca-Cola Singapore plant general manager Gerardo Artavia told The Straits Times it is working with A*Star to build an analytical AI system to help the company distribute beverages to its markets more efficiently.

“We need to know how much to sell in each market, and not send too many bottles that they become expired,” he said. “AI can help us to find a formula to find that balance.”

Mr Artavia said that an initiative like the one with A*Star at AIMfg will go a long way to help manufacturing firms adopt AI, as many may lack the funding and expertise to digitalise their services.

 

[syed putra]

They outsource AI software engineers from India?

 

[k1976]

They outsource AI software engineers from India?

U guess correct de woh

 

[k1976]

Use bayi...no worry of leaking important trade know how to cina mah

 

[eatshitndie]

i asked ai to shit on my behalf but it could not as it replied “no shit!”