(Bloomberg) — Nvidia Corp., whose powerful chips helped set the stage for the artificial intelligence boom, is now looking to address a major concern surrounding the technology: that AI bots will go rogue and cause harm.
The company is introducing software Tuesday that regulates AI systems based on large language models — the learning technique used by OpenAI’s ChatGPT and other popular bots. The tool, called NeMo Guardrails, can keep chatbots on topic and make them less likely to offer up restricted information. It also will prevent them from guessing wrongly or taking actions outside their purview, Nvidia said.
The flood of interest in ChatGPT, along with other systems that mine large data sets to generate automated answers, has the potential to change almost every industry. The trend is also poised to make a fortune for Nvidia, a pioneer of graphics cards that now gets most of its money from chips for data centers — the server farms needed to power AI. But for artificial intelligence to continue to flourish, users need to be able to trust the results that chatbots generate.
“Everyone is aware of the power of generative large language models,” said Jonathan Cohen, a vice president of applied research at Nvidia. “It’s important that they’re deployed in a way that’s safe and secure.”
Some of the largest companies in technology use Nvidia’s processors to handle AI work within data centers, and that’s helped the chipmaker weather a broader slump in the computer industry. In fact, its data center unit is now bigger than the whole company was as recently as 2020.
Nvidia is offering NeMo Guardrails as open-source software and will continue to update it. The Santa Clara, California-based company is also including it in a suite of programs that it provides to clients for a fee.
NeMo Guardrails will operate as a layer between the end user and the AI program. Using a mixture of Nvidia’s own large-language model and conventional software, the system will be able to recognize when the user is asking a factual question and check whether the bot can and should answer that query. It will determine whether the answer generated is based on facts and govern the manner in which that chatbot gives a reply.
For example, let’s say an employee asks a human-resources chatbot if the company provides support for workers who want to adopt a child. That would pass through NeMo Guardrails and return text containing the company’s related benefits. Asking the same bot how many employees have taken advantage of that benefit would trigger a refusal to comply, because the data is confidential.
If users asked the bot for nonpublic financial information on the company, they’d be told it was off topic. And to check whether the program really knows the answer and isn’t just guessing – a problem known as hallucination – the software asks the question multiple times in the background to ensure that the user isn’t getting a random but superficially plausible response. Similarly, the software might make sure that a bot stays measured in the way it reacts, even when users attempt to provoke it into replying in a way that’s inappropriate.
In a recent controversy, ChatGPT users described gaining access to forbidden information by asking the bot to pretend it was their dead grandmother.
Making Nvidia’s new tool free to access will let the community test it out and help make sure that it can’t be used for further abuse, Cohen said.
“Any time you open-source something, people can examine it and find a way to exploit it. That’s why we open-sourced it,” he said. “We want the community to look at it.”
A handy little update to the official ChatGPT app means you can now use Apple’s Siri to talk to the chatbot.
The useful trick works through an Apple feature called Shortcuts that lets you automate tasks on your iPhone.
Once set up, you can basically ask Siri to fetch answers from ChatGPT. It’s kind of like rewiring Siri’s brain to make the otherwise limited digital assistant a lot smarter. Until Apple gives the voice bot an AI language upgrade, this is the next best thing.
OpenAI launched the free ChatGPT iPhone app in the US last month, and quickly brought it to the UK. Although the app already offers a voice input feature that relies on its Whisper speech-recognition system, linking it to Siri means you can also use it on other Apple devices like a HomePod smart speaker.
Setting up the new feature is a breeze – here’s what you need to do.
Now, head to the Shortcuts app, which is available on any iPhone running iOS 12 or later.
Here, you can click the plus icon in the top-right corner and search for ChatGPT to set up the default Shortcut.
However, if you want to customise it, then you should click on the Shortcuts button in the top-left corner, and you should see ChatGPT under “App Shortcuts.”
Now, select “Add to shortcut” and tweak it how you want. For instance, you can set it to react to a specific command, such as “Hey Siri, Ask AI” instead of the standard prompt “Hey Siri, Ask ChatGPT”.
Watch out for ChatGPT clones
Within three weeks of its launch, the OpenAI ChatGPT app was downloaded 5 million times, according to data.ai.
The bot landed in an App Store full of clones and duplicates with similar-sounding names, some of which were using shady tactics to charge users for their services.
Thankfully, you can usually spot these dubious “fleeceware” apps by their low review scores. Google and Apple also removed some of the worst rip-offs after they were contacted by cybersecurity researchers.
Assassin’s Creed Mirage finally showed the much-anticipated return of classic elements from the franchise during the PlayStation Showcase. This trailer highlighted various elements, encouraging fans to anticipate future gameplay from Ubisoft.
A developer diary from the studio has now been released, highlighting classic elements featured in the Assassin’s Creed title. This includes stealth, assassinations, and parkour mechanics.
Why it matters: Many fans have been waiting for a classic Assassin’s Creed experience since the RPG trilogy. Mirage promises a return to the basic formula, focusing on the core Assassin’s Creed experience.
Here is the full version of #AssassinsCreed Mirage – “A Return to the Roots”!
This is the first of series of 3 dev diaries on #AssassinsCreedMirage and is specifically focused on Parkour, Stealth and Assassinations!
— Access The Animus (@AccessTheAnimus) June 9, 2023
The developer diary explains Ubisoft’s thought process and focuses when returning to the classic formula.
Assassin’s Creed Mirage brings parkour mechanics like elevators, vaulting, and a dense map that encourages creative movement. Basim is more agile than recent Assassins, making parkour more fluid than before. Ubisoft also introduces new mechanics like the pole vault.
For fans of stealth, elements like social stealth will be making a comeback. Additionally, players can utilize bench assassinations, and Ubisoft has added more animations for stealth kills. This should ensure players are encouraged to pick the stealthy approach as they play through the upcoming release.
The game promotes freedom in story missions, allowing players to approach assassinations at their own pace. Assassin’s Creed Mirage also features bureaus that act as mission hubs, allowing players to take up side quests like assassinations and rescue missions.
Latest Updates
Assassin’s Creed Mirage is set to launch on October 12. Ubisoft has previously stated that the game was inspired by fan demand for a classic Assassin’s Creed experience, and the first developer diary shows just a few ways, in which Ubisoft is looking to bring the old formula to life.
One interpretation of the beginning footage in the trailer is that it’s showing the events in an alternate timeline. Cloud is missing from the group of Avalanche members’ bodies recovered by Shinra, seemingly after their attempt to break out of the headquarters or potentially during a routine reactor bombing mission. The fact that the spikey-haired hero is missing has some fans wondering if the footage is a brief look at a Final Fantasy VII timeline where he never made it to Midgar, or perhaps one where he joins Sephiroth instead of fighting him at the end of Final Fantasy VII Remake.
Another theory is that the footage is made-up altogether. We don’t actually see direct shots of the Avalanche members’ bodies. Instead it’s through the eyes of Shinra TV, an organization with a clear and documented history of fake news and propaganda. It’s possible the footage is just showing how Shinra will eventually try to cover up the team’s headquarters rampage and ensuing car chase to keep the residents of Midgar calm and in the dark.
The trailer gives us our first look at a Chocobo out in the wild at the farm where players were eventually able to breed them in the original game. It’s unclear if that mechanic will be saved for the final game in the remake trilogy, or if players will get the chance to breed a Gold Chocobo in Rebirth and even take it to race it at the Gold Saucer. I can only imagine how many hours I’ll be spending on the Chocobo tracks if that’s the case.
The Turks are on the chase again and Elena is their newest member. The reckless recruit is as much of a punk as ever in the new trailer. She has a crush on Tseng, the leader of the Turks, and her antics with Rude and the rest will no doubt spawn an endless number of memes. Her cameo in the trailer is already leading to a ton of fan art.
Ho Ho Hooo! Bugenhagen has made the jump to PS5 as well and the Cosmo Canyon elder looks glorious. Red XIII’s surrogate grandfather can be heard explaining how the planet relies on the Lifestream to exist, and he still doesn’t have legs and floats around instead. One of the game’s most beloved and strange-looking characters in his original 32-bit form, the glow up is real and players are loving it.
The Shinra airship doesn’t get a moment to shine in the trailer, but it can be seen in the background of the initial shot of the seaside military base Juno. It might end up that only in the trilogy’s final game will players get to step inside and take the reins over for themselves, but in the meantime the Highwind is already getting fans excited for when that moment comes.
“I saw you lying there. I figured it was too late,” Cloud said in the first Rebirth trailer. “Wait, what are you implying? That I died? That I’m some kind of imposter?” responded Tifa. In the new trailer Sephiroth says, “You know that I killed her, so who is she?” This has fans wondering if the Tifa we know from Remake is actually a clone, or a version of Jenova, Sephiroth’s mutant alien half-mom. It’s possible that the developers are just building intrigue by how the dialogue is framed, or maybe it’s laying the groundwork for a radically different outcome to the rest of Final Fantasy VII Rebirth’s story.
This all culminates in what would be the weirdest twist of all: Aerith not dying in the remake trilogy. The dream of saving Final Fantasy VII’s iconic flower girl is as old as the game itself, with urban legends that an option to stop Sephiroth from killing her or bring her back to life were hidden inside the game or cut at the last minute. If Tifa, or at least the original version of her, was killed at some point, maybe it implies that Aerith will live instead, or that Cloud will have the option to choose between different timelines where one set of tragic events is replaced with another. Or maybe this is all just the same type of wishful thinking Final Fantasy VII fans have been using to cope since the game came out. We’ll see what answers Rebirth provides when it arrives early next year. For now it’s fun to believe anything’s possible.
Kotaku is covering everything Summer Game Fest, from the main show on Thursday to other events happening throughout the next week. Whether you’re into larger-than-life triple-A games or intimate, offbeat indies, you can keep up with all things SGF here.
