Home Learn Why Big Tech’s bet on AI assistants is so dangerous

Why Big Tech’s bet on AI assistants is so dangerous

0
Why Big Tech’s bet on AI assistants is so dangerous

For the reason that starting of the generative AI boom, tech firms have been feverishly attempting to give you the killer app for the technology. First it was online search, with mixed results. Now it’s AI assistants. Last week, OpenAI, Meta, and Google launched latest features for his or her AI chatbots that allow them to go looking the online and act as a form of non-public assistant. 

OpenAI unveiled latest ChatGPT features that include the flexibility to have a conversation with the chatbot as when you were making a call, allowing you to immediately get responses to your spoken questions in a lifelike synthetic voice, as my colleague Will Douglas Heaven reported. OpenAI also revealed that ChatGPT will give you the chance to search the online.  

Google’s rival bot, Bard, is plugged into many of the company’s ecosystem, including Gmail, Docs, YouTube, and Maps. The concept is that folks will give you the chance to make use of the chatbot to ask questions on their very own content—for instance, by getting it to go looking through their emails or organize their calendar. Bard can even give you the chance to immediately retrieve information from Google Search. In an identical vein, Meta too announced that it’s throwing AI chatbots at every thing. Users will give you the chance to ask AI chatbots and celebrity AI avatars questions on WhatsApp, Messenger, and Instagram, with the AI model retrieving information online from Bing search. 

This can be a dangerous bet, given the constraints of the technology. Tech firms haven’t solved a few of the persistent problems with AI language models, reminiscent of their propensity to make things up or “hallucinate.” But what concerns me probably the most is that they’re a security and privacy disaster, as I wrote earlier this 12 months. Tech firms are putting this deeply flawed tech within the hands of thousands and thousands of individuals and allowing AI models access to sensitive information reminiscent of their emails, calendars, and personal messages. In doing so, they’re making us all vulnerable to scams, phishing, and hacks on an enormous scale. 

I’ve covered the numerous security problems with AI language models before. Now that AI assistants have access to private information and may concurrently browse the online, they’re particularly liable to a style of attack called indirect prompt injection. It’s ridiculously easy to execute, and there isn’t a known fix. 

In an indirect prompt injection attack, a 3rd party “alters a web site by adding hidden text that is supposed to vary the AI’s behavior,” as I wrote in April. “Attackers could use social media or email to direct users to web sites with these secret prompts. Once that happens, the AI system might be manipulated to let the attacker attempt to extract people’s bank card information, for instance.” With this latest generation of AI models plugged into social media and emails, the opportunities for hackers are limitless. 

I asked OpenAI, Google, and Meta what they’re doing to defend against prompt injection attacks and hallucinations. Meta didn’t reply in time for publication, and OpenAI didn’t comment on the record. 

Regarding AI’s propensity to make things up, a spokesperson for Google did say the corporate was releasing Bard as an “experiment,” and that it lets users fact-check Bard’s answers using Google Search. “If users see a hallucination or something that isn’t accurate, we encourage them to click the thumbs-down button and supply feedback. That’s a method Bard will learn and improve,” the spokesperson said. In fact, this approach puts the onus on the user to identify the error, and folks tend to position an excessive amount of trust within the responses generated by a pc. Google didn’t have a solution for my query about prompt injection. 

For prompt injection, Google confirmed it isn’t a solved problem and stays an lively area of research. The spokesperson said the corporate is using other systems, reminiscent of spam filters, to discover and filter out attempted attacks, and is conducting adversarial testing and red teaming exercises to discover how malicious actors might attack products built on language models. “We’re using specially trained models to assist discover known malicious inputs and known unsafe outputs that violate our policies,” the spokesperson said.  

Now, I get that there’ll all the time be early teething pains with every latest product launch. But it surely’s saying quite a bit when even early cheerleaders of AI language model products have not been that impressed. Kevin Roose, a Latest York Times columnist, found that Google’s assistant was good at summarizing emails but in addition told him about emails that weren’t in his inbox. 

TL;DR? Tech firms shouldn’t be so complacent concerning the purported “inevitability” of AI tools. Bizarre people don’t are inclined to adopt technologies that keep failing in annoying and unpredictable ways, and it’s only a matter of time until we see the hackers using these latest AI assistants maliciously. Straight away, we’re all sitting geese. 

I don’t find out about you, but I intend to attend a bit of longer before letting this generation of AI systems snoop around in my email. 

Deeper Learning

This robotic exoskeleton may help runners sprint faster

Now that is cool. An exoskeleton may help runners increase their speed by encouraging them to take more steps, allowing them to cover short distances more quickly. A team of researchers at Chung-Ang University in Seoul, South Korea, built a light-weight exosuit that helps people run faster by assisting their hip extension—the powerful motion that propels a runner forward. The suit’s sensors feed data into algorithms that track each runner’s individual running style and speed.

Harder, higher, faster, stronger: The team tested the exosuit on nine young male runners, none of whom were considered to be elite athletes. The lads sprinted outside in a straight line for 200 meters twice, once wearing the exosuit and once without. On average, the participants managed to run the gap 0.97 seconds faster after they were wearing the suit than after they weren’t.  Read more from Rhiannon Williams here. 

Bits and Bytes

Hollywood writers and studios reached a deal on the usage of AI
The Writers Guild of America and the Alliance of Motion Picture and Television Producers have reached a deal ending the Hollywood writers’ strike and agreeing on terms of use for AI. The deal  stipulates that AI systems can’t be used to write down or rewrite any scripts, and that studios must disclose when they provide writers AI-generated materials. Writers can even give you the chance to make a decision whether their scripts are used to coach AI models. This move ensures that folks can use AI as a tool, moderately than simply being replaced by it. (Wired)

A French AI company launched an AI chatbot that provides detailed instructions on murder and ethnic cleansing
Eugh. Mistral, a French startup founded by former Meta and DeepMind people, has launched an open-source AI language model, which outperforms Meta’sLlama in some metrics. But unlike Llama, Mistral has no content filters and spews toxic content with no restrictions. (404 Media) 

OpenAI is planning for a consumer device
OpenAI is in “advanced talks” with former Apple designer Sir Jony Ive and SoftBank to construct the “iPhone of artificial intelligence.” It’s unclear what that device would appear to be or do. Consumer hardware is difficult to get right. Many tech firms have announced—and scrapped—ambitious plans to roll out consumer tech. My money is on a voice-controlled AI assistant. (The Information) 

These 183,000 books are fueling the most important fight in publishing and tech
It will turn out to be useful in copyright lawsuits against tech firms—a searchable database that enables you to see which books and authors have been scraped into data sets to coach generative AI systems. (The Atlantic) 

LEAVE A REPLY

Please enter your comment!
Please enter your name here