Google I/O 2023 has showcased the company’s efforts to integrate AI capabilities from PaLM 2 across the Google ecosystem, including Bard. However, Google has also expressed concern about potential tensions between misinformation and certain AI capabilities, particularly those behind deepfakes. To prevent the misuse of these features, Google is integrating artefacts into photos and videos, such as watermarks and metadata.
One new tool showcased at the event is Google’s universal translator, an experimental AI video dubbing service that translates speech in real-time. The model works in four stages, matching lip movements in a video to recognised words, providing instant speech generation, using intonation to aid translation and finally replicating the style and tone to generate the translation.
Although the universal translator feature isn’t yet available outside of a small testing group, it could be rolled out to services such as YouTube and Google Meet once Google has tested numerous safeguards. Being able to translate live videos in real-time into multiple languages could be an incredibly useful tool, expanding a YouTube channel’s global viewership and allowing for more collaborative projects across countries.
Google’s efforts to integrate AI capabilities across its ecosystem are impressive, but the company also recognizes the potential for misuse. By implementing safeguards and guardrails, Google is taking steps to prevent the misuse of these features and ensure that AI technology is used ethically and responsibly. We look forward to seeing where this technology goes in the future and how it will shape our interactions with the world.
