AI

Are AI models doomed to always hallucinate?

Comment

Text to video concept, text-to-video by generative AI. Language model technology. Cyborg hand holding vdo generated by artificial intelligence.
Image Credits: Ole_CNX (opens in a new window) / Getty Images

Large language models (LLMs) like OpenAI’s ChatGPT all suffer from the same problem: they make stuff up.

The mistakes range from strange and innocuous — like claiming that the Golden Gate Bridge was transported across Egypt in 2016 — to highly problematic, even dangerous.

A mayor in Australia recently threatened to sue OpenAI because ChatGPT mistakenly claimed he pleaded guilty in a major bribery scandal. Researchers have found that LLM hallucinations can be exploited to distribute malicious code packages to unsuspecting software developers. And LLMs frequently give bad mental health and medical advice, like that wine consumption can “prevent cancer.”

This tendency to invent “facts” is a phenomenon known as hallucination, and it happens because of the way today’s LLMs — and all generative AI models, for that matter — are developed and trained.

Training models

Generative AI models have no real intelligence — they’re statistical systems that predict words, images, speech, music or other data. Fed an enormous number of examples, usually sourced from the public web, AI models learn how likely data is to occur based on patterns, including the context of any surrounding data.

For example, given a typical email ending in the fragment “Looking forward…”, an LLM might complete it with “… to hearing back” — following the pattern of the countless emails it’s been trained on. It doesn’t mean the LLM is looking forward to anything.

“The current framework of training LLMs involves concealing, or ‘masking,’ previous words for context” and having the model predict which word should follow this context, Sebastian Berns, a Ph.D. researchers at Queen Mary University of London, told TechCrunch in an email interview. “This is conceptually similar to using predictive text in iOS and continually pressing one of the suggested next words.”

This probability-based approach works remarkably well at scale — for the most part. But while the range of words and their probabilities are likely to result in text that makes sense, it’s far from certain.

The emerging types of language models and why they matter

LLMs can generate something that’s grammatically correct but nonsensical, for instance — like the claim about the Golden Gate. Or they can spout mistruths, propagating inaccuracies in their training data. Or they can conflate different sources of information, including fictional sources, even if those sources clearly contradict each other.

It’s not malicious on the LLMs’ part. They don’t have malice, and the concepts of true and false are meaningless to them. They’ve simply learned to associate certain words or phrases with certain concepts, even if those associations aren’t accurate.

“‘Hallucinations’ are connected to the inability of an LLM to estimate the uncertainty of its own prediction,” Berns said. “An LLM is typically trained to always produce an output, even when the input is very different from the training data. A standard LLM does not have any way of knowing if it’s capable of reliably answering a query or making a prediction.”

Solving hallucination

The question is, can hallucination be solved? It depends on what you mean by “solved.”

Vu Ha, an applied researcher and engineer at the Allen Institute for Artificial Intelligence, asserts that LLMs “do and will always hallucinate.” But he also believes there are concrete ways to reduce — albeit not eliminate — hallucinations, depending on how an LLM is trained and deployed. 

“Consider a question answering system,” Ha said via email. “It’s possible to engineer it to have high accuracy by curating a high-quality knowledge base of questions and answers, and connecting this knowledge base with an LLM to provide accurate answers via a retrieval-like process.”

Ha illustrated the difference between an LLM with a “high-quality” knowledge base to draw on versus one with less careful data curation. He ran the question “Who are the authors of the Toolformer paper?” (Toolformer is an AI model trained by Meta) through Microsoft’s LLM-powered Bing Chat and Google’s Bard. Bing Chat correctly listed all eight Meta co-authors, while Bard misattributed the paper to researchers at Google and Hugging Face.

“Any deployed LLM-based system will hallucinate. The real question is if the benefits outweigh the negative outcome caused by hallucination,” Ha said. In other words, if there’s no obvious harm done by a model — the model gets a date or name wrong once in a while, say — but it’s otherwise helpful, then it might be worth the trade-off. “It’s a question of maximizing expected utility of the AI,” he added.

Age of AI: Everything you need to know about artificial intelligence

Berns pointed out another technique that had been used with some success to reduce hallucinations in LLMs: reinforcement learning from human feedback (RLHF). Introduced by OpenAI in 2017, RLHF involves training an LLM, then gathering additional information to train a “reward” model and fine-tuning the LLM with the reward model via reinforcement learning.

In RLHF, a set of prompts from a predefined dataset are passed through an LLM to generate new text. Then, human annotators are used to rank the outputs from the LLM in terms of their overall “helpfulness” — data that’s used to train the reward model. The reward model, which at this point can take in any text and assign it a score of how well humans perceive it, is then used to fine-tune the LLM’s generated responses.

OpenAI leveraged RLHF to train several of its models, including GPT-4. But even RLHF isn’t perfect, Berns warned.

“I believe the space of possibilities is too large to fully ‘align’ LLMs with RLHF,” Berns said. “Something often done in the RLHF setting is training a model to produce an ‘I don’t know’ answer [to a tricky question], primarily relying on human domain knowledge and hoping the model generalizes it to its own domain knowledge. Often it does, but it can be a bit finicky.”

Alternative philosophies

Assuming hallucination isn’t solvable, at least not with today’s LLMs, is that a bad thing? Berns doesn’t think so, actually. Hallucinating models could fuel creativity by acting as a “co-creative partner,” he posits — giving outputs that might not be wholly factual but that contain some useful threads to tug on nonetheless. Creative uses of hallucination can produce outcomes or combinations of ideas that might not occur to most people.

“‘Hallucinations’ are a problem if generated statements are factually incorrect or violate any general human, social or specific cultural values — in scenarios where a person relies on the LLM to be an expert,” he said. “But in creative or artistic tasks, the ability to come up with unexpected outputs can be valuable. A human recipient might be surprised by a response to a query and therefore be pushed into a certain direction of thoughts which might lead to the novel connection of ideas.”

Ha argued that the LLMs of today are being held to an unreasonable standard — humans “hallucinate” too, after all, when we misremember or otherwise misrepresent the truth. But with LLMs, he believes we experience a cognitive dissonance because the models produce outputs that look good on the surface but contain errors upon further inspection.

“Simply put, LLMs, just like any AI techniques, are imperfect and thus make mistakes,” he said. “Traditionally, we’re OK with AI systems making mistakes since we expect and accept imperfections. But it’s more nuanced when LLMs make mistakes.”

Indeed, the answer may well not lie in how generative AI models work at the technical level. Insofar as there’s a “solution” to hallucination today, treating models’ predictions with a skeptical eye seems to be the best approach.

More TechCrunch

Some Indian government websites have allowed scammers to plant advertisements capable of redirecting visitors to online betting platforms. TechCrunch discovered around four dozen “gov.in” website links associated with Indian states,…

Scammers found planting online betting ads on Indian government websites

Around 550 employees across autonomous vehicle company Motional have been laid off, according to information taken from WARN notice filings and sources at the company.  Earlier this week, TechCrunch reported…

Motional cut about 550 employees, around 40%, in recent restructuring, sources say

The deck included some redacted numbers, but there was still enough data to get a good picture.

Pitch Deck Teardown: Cloudsmith’s $15M Series A deck

The company is describing the event as “a chance to demo some ChatGPT and GPT-4 updates.”

OpenAI’s ChatGPT announcement: What we know so far

Unlike ChatGPT, Claude did not become a new App Store hit.

Anthropic’s Claude sees tepid reception on iOS compared with ChatGPT’s debut

Welcome to Startups Weekly — Haje‘s weekly recap of everything you can’t miss from the world of startups. Sign up here to get it in your inbox every Friday. Look,…

Startups Weekly: Trouble in EV land and Peloton is circling the drain

Scarcely five months after its founding, hard tech startup Layup Parts has landed a $9 million round of financing led by Founders Fund to transform composites manufacturing. Lux Capital and Haystack…

Founders Fund leads financing of composites startup Layup Parts

AI startup Anthropic is changing its policies to allow minors to use its generative AI systems — in certain circumstances, at least.  Announced in a post on the company’s official…

Anthropic now lets kids use its AI tech — within limits

Zeekr’s market hype is noteworthy and may indicate that investors see value in the high-quality, low-price offerings of Chinese automakers.

The buzziest EV IPO of the year is a Chinese automaker

Venture capital has been hit hard by souring macroeconomic conditions over the past few years and it’s not yet clear how the market downturn affected VC fund performance. But recent…

VC fund performance is down sharply — but it may have already hit its lowest point

The person who claims to have 49 million Dell customer records told TechCrunch that he brute-forced an online company portal and scraped customer data, including physical addresses, directly from Dell’s…

Threat actor says he scraped 49M Dell customer addresses before the company found out

The social network has announced an updated version of its app that lets you offer feedback about its algorithmic feed so you can better customize it.

Bluesky now lets you personalize main Discover feed using new controls

Microsoft will launch its own mobile game store in July, the company announced at the Bloomberg Technology Summit on Thursday. Xbox president Sarah Bond shared that the company plans to…

Microsoft is launching its mobile game store in July

Smart ring maker Oura is launching two new features focused on heart health, the company announced on Friday. The first claims to help users get an idea of their cardiovascular…

Oura launches two new heart health features

Keeping up with an industry as fast-moving as AI is a tall order. So until an AI can do it for you, here’s a handy roundup of recent stories in the world…

This Week in AI: OpenAI considers allowing AI porn

Garena is quietly developing new India-themed games even though Free Fire, its biggest title, has still not made a comeback to the country.

Garena is quietly making India-themed games even as Free Fire’s relaunch remains doubtful

The U.S.’ NHTSA has opened a fourth investigation into the Fisker Ocean SUV, spurred by multiple claims of “inadvertent Automatic Emergency Braking.”

Fisker Ocean faces fourth federal safety probe

CoreWeave has formally opened an office in London that will serve as its European headquarters and home to two new data centers.

CoreWeave, a $19B AI compute provider, opens European HQ in London with plans for 2 UK data centers

The Series C funding, which brings its total raise to around $95 million, will go toward mass production of the startup’s inaugural products

AI chip startup DEEPX secures $80M Series C at a $529M valuation 

A dust-up between Evolve Bank & Trust, Mercury and Synapse has led TabaPay to abandon its acquisition plans of troubled banking-as-a-service startup Synapse.

Infighting among fintech players has caused TabaPay to ‘pull out’ from buying bankrupt Synapse

The problem is not the media, but the message.

Apple’s ‘Crush’ ad is disgusting

The Twitter for Android client was “a demo app that Google had created and gave to us,” says Particle co-founder and ex-Twitter employee Sara Beykpour.

Google built some of the first social apps for Android, including Twitter and others

WhatsApp is updating its mobile apps for a fresh and more streamlined look, while also introducing a new “darker dark mode,” the company announced on Thursday. The messaging app says…

WhatsApp’s latest update streamlines navigation and adds a ‘darker dark mode’

Plinky lets you solve the problem of saving and organizing links from anywhere with a focus on simplicity and customization.

Plinky is an app for you to collect and organize links easily

The keynote kicks off at 10 a.m. PT on Tuesday and will offer glimpses into the latest versions of Android, Wear OS and Android TV.

Google I/O 2024: How to watch

For cancer patients, medicines administered in clinical trials can help save or extend lives. But despite thousands of trials in the United States each year, only 3% to 5% of…

Triomics raises $15M Series A to automate cancer clinical trials matching

Welcome back to TechCrunch Mobility — your central hub for news and insights on the future of transportation. Sign up here for free — just click TechCrunch Mobility! Tap, tap.…

Tesla drives Luminar lidar sales and Motional pauses robotaxi plans

The newly announced “Public Content Policy” will now join Reddit’s existing privacy policy and content policy to guide how Reddit’s data is being accessed and used by commercial entities and…

Reddit locks down its public data in new content policy, says use now requires a contract

Eva Ho plans to step away from her position as general partner at Fika Ventures, the Los Angeles-based seed firm she co-founded in 2016. Fika told LPs of Ho’s intention…

Fika Ventures co-founder Eva Ho will step back from the firm after its current fund is deployed

In a post on Werner Vogels’ personal blog, he details Distill, an open-source app he built to transcribe and summarize conference calls.

Amazon’s CTO built a meeting-summarizing app for some reason