Fun

News Feed - 2023-10-03 02:10:00

Tristan Greene6 hours agoResearchers find LLMs like ChatGPT output sensitive data even after it’s been ‘deleted’According to the scientists, there’s no universal method by which data can be deleted from a pretrained large language model.2784 Total views12 Total sharesListen to article 0:00NewsJoin us on social networksA trio of scientists from the University of North Carolina, Chapel Hill recently published preprint artificial intelligence (AI) research showcasing how difficult it is to remove sensitive data from large language models (LLMs) such as OpenAI’s ChatGPT and Google’s Bard. 


According to the researchers" paper, the task of “deleting” information from LLMs is possible, but it’s just as difficult to verify the information has been removed as it is to actually remove it.


The reason for this has to do with how LLMs are engineered and trained. The models are pretrained on databases and then fine-tuned to generate coherent outputs (GPT stands for “generative pretrained transformer”).


Once a model is trained, its creators cannot, for example, go back into the database and delete specific files in order to prohibit the model from outputting related results. Essentially, all the information a model is trained on exists somewhere inside its weights and parameters where they’re undefinable without actually generating outputs. This is the “black box” of AI.


A problem arises when LLMs trained on massive datasets output sensitive information such as personally identifiable information, financial records, or other potentially harmful and unwanted outputs.


Related:Microsoft to form nuclear power team to support AI: Report


In a hypothetical situation where an LLM was trained on sensitive banking information, for example, there’s typically no way for the AI’s creator to find those files and delete them. Instead, AI devs use guardrails such as hard-coded prompts that inhibit specific behaviors or reinforcement learning from human feedback (RLHF).


In an RLHF paradigm, human assessors engage models with the purpose of eliciting both wanted and unwanted behaviors. When the models’ outputs are desirable, they receive feedback that tunes the model toward that behavior. And when outputs demonstrate unwanted behavior, they receive feedback designed to limit such behavior in future outputs.Despite being “deleted” from a model"s weights, the word “Spain” can still be conjured using reworded prompts. Image source: Patil, et. al., 2023


However, as the UNC researchers point out, this method relies on humans finding all the flaws a model might exhibit, and even when successful, it still doesn’t “delete” the information from the model.


Per the team’s research paper:“A possibly deeper shortcoming of RLHF is that a model may still know the sensitive information. While there is much debate about what models truly ‘know’ it seems problematic for a model to, e.g., be able to describe how to make a bioweapon but merely refrain from answering questions about how to do this.”


Ultimately, the UNC researchers concluded that even state-of-the-art model editing methods, such as Rank-One Model Editing “fail to fully delete factual information from LLMs, as facts can still be extracted 38% of the time by whitebox attacks and 29% of the time by blackbox attacks.”


The model the team used to conduct their research is called GPT-J. While GPT-3.5, one of the base models that power ChatGPT, was fine-tuned with 170 billion parameters, GPT-J only has 6 billion.


Ostensibly, this means the problem of finding and eliminating unwanted data in an LLM such as GPT-3.5 is exponentially more difficult than doing so in a smaller model.


The researchers were able to develop new defense methods to protect LLMs from some “extraction attacks” — purposeful attempts by bad actors to use prompting to circumvent a model’s guardrails in order to make it output sensitive information


However, as the researchers write, “the problem of deleting sensitive information may be one where defense methods are always playing catch-up to new attack methods.”# AI# Machine Learning# ChatGPTAdd reactionAdd reactionRead moreHow to use index funds and ETFs for passive crypto incomeAI tech boom: Is the artificial intelligence market already saturated?AI a powerful tool for devs to change gaming, says former Google gaming head

News Feed

Prashant Jha13 hours agoAbracadabra proposes hiking loan interest rate by 200% to manage Curve riskThe Curve Finance exploit has created a liquidity crisis in the DeFi ecosystem, with several lending protocols rushing to
Airbnb Looking to Support Crypto While Focusing on Free Housing for 100,000 Ukrainian Refugees
Airbnb Looking to Support Crypto While Focusing on Free Housing for 100,000 Ukrainian Refugees Airbnb confirms that it is looking to support cryptocurrency. The company has also of
Robert Kiyosaki Warns of ‘Biggest Bond Crash Since 1788’ — Waiting to Buy Bitcoin at Lower Price
Robert Kiyosaki Warns of "Biggest Bond Crash Since 1788" — Waiting to Buy Bitcoin at Lower Price The famous author of the best-selling book Rich Dad Poor Dad, Robert Kiyosaki, ha
5 Altcoins to Buy Now as Bitcoin Price Hovers at $95K After DOJ Dip
Although Bitcoin’s recent price drop led many to wonder whether this is the end of its euphoria. However, as it’s looking to turn out, all BTC really did was go into healthy consolidation after a big move in
More crypto AI alliances emerge following $7.5B token merger
Zhiyuan Sun6 hours agoMore crypto AI alliances emerge following $7.5B token mergerMany crypto projects have integrated AI as part of their everyday routine since the introduction of LLMs last year.1300 Total views7 Total
Bitcoin miners may gain $13.9B yearly from 20% shift to AI and HPC: VanEck
Ciaran Lyons5 hours agoBitcoin miners may gain $13.9B yearly from 20% shift to AI and HPC: VanEckBitcoin miners could increase profitability and improve “bad balance sheets” by allocating some of their energy capacit
Microstrategy to Sell $600 Million Worth of Convertible Notes to Buy More Bitcoin
Microstrategy to Sell $600 Million Worth of Convertible Notes to Buy More Bitcoin The publicly listed company Microstrategy announced that it has plans to purcha
Latam Still Unprepared to Deal With Crypto Crime and Scams, According to GFI Report
Latam Still Unprepared to Deal With Crypto Crime and Scams, According to GFI Report Latam is still unprepared to deal with cryptocurrency-related crimes and scam situations, accord
$RMPL Launches, Revolutionizing the Cryptocurrency Market with a Decentralized Elastic Supply Model
$RMPL Launches, Revolutionizing the Cryptocurrency Market with a Decentralized Elastic Supply ModelAnnouncing the official launch of $RMPL, an ERC20 cryptocurrency with an elastic s
Trump Views Crypto a Threat, Proposes Countermeasures in New Budget
Trump Views Crypto a Threat, Proposes Countermeasures in New Budget The White House released the Budget of the U.S. Government on Monday, outlining some proposals to increase cry
Analyst Puts XRP Cycle High At $20-$30, Here’s Why
Reason to trust Strict editorial policy that focuses on accuracy, relevance, and impartiality Created by industry experts and meticulously reviewed The highest standards in reporting and pu
Brian Quarmby2 hours agoDemand for Bitcoin could grow by up to 10X within 12 months: Michael SaylorThe MicroStrategy co-founder emphasized that demand for Bitcoin will significantly ramp up following the halving next yea