In the run up to introducing its own AI services, search behemoth Google is leaving no stones unturned. On Friday, it was reported that the Alphabet Inc. company has invested over $400 million in artificial intelligence startup Anthropic, putting it in the spotlight. Anthropic has been testing a potential rival to OpenAI’s revolutionary chat bot ChatGPT. And, this move comes at a time when Google, which has been feeling the heat, has been making stealthy moves to ace its AI game.
Anthropic was founded by Dario Amodei, former VP of research at OpenAI and his sister Daniela in 2021 in San Francisco with the aim to create large-scale AI systems that are interpretable and robust. According to a report published in TechCrunch, Dario moved out of OpenAI owing to his belief that there was a need for the AI industry to move beyond a handful of innovations such as GPT-3. The Amodei siblings have been working on the core idea that most of the AI models out there are not fully understood.
In January, it came to the attention of the AI community that the start-up founded by OpenAI’s former employees, developed a similar AI system to ChatGPT known as Claude. Since it was available as a part of closed beta, not much has been known about it. Regardless, several testers in the beta have detailed their interactions with Claude on social media.
Claude explained
Anthropic created Claude using a method known as ‘constitutional AI’. The company in a thread shared on Twitter on December 16 last year, said that Claude aims to offer a ‘principle-based’ approach that aligns AI systems with human intentions. This essentially means that it uses AI similar to ChatGPT to respond to questions based on a simple set of principles.
In order to build Claude, Anthropic began by assembling a list of 10 principles that combine to form a ‘constitution’. While these principles are not in the public domain yet, Anthropic has revealed that they are backed by the concepts of beneficence which is maximising positive impact; nonmaleficence – not giving harmful advice, and autonomy which stands for respecting the freedom of choice. In essence, Claude is similar to ChatGPT and other language models as it is a statistical tool that predicts words.
According to Anthropic, Claude consumed a vast set of text examples from the web and learned how words occur based on the semantic context surrounding text. This has enabled Claude to have open-ended conversations, crack jokes, and engage in philosophical debate on a wide range of topics.
What does all this mean to Google?
Both Google and Anthropic have not commented on the investment. However, they have separately announced a partnership where the AI start-up will be using Google’s cloud computing services. The investment comes at a time when Google’s parent company missed both top and bottom lines in the fourth quarter, and the widespread popularity of ChatGPT. The emergence of ChatGPT and numerous other AI-backed technologies in recent times seems to have pushed the search engine giant to devise newer ways to stay on top of the game.
The rise of OpenAI’s chatbot is looming large on the search giant. While a routine search would render links and images to respective queries, ChatGPT yields concise human-like sentences in response. The application goes a notch higher with its ability to create content from scratch which is certainly seen as disruptive. All these certainly are signaling a time for major upheaval for Google and its line of products and services.
Reinventing through AI
Perhaps, it’s for this reason that Google seems to be scrambling to come out with its own AI innovation. A recent report from The New York Times stated that the tech behemoth has sought the help of Larry Page and Sergey Brin, the founders of its parent company Alphabet. The same report also said that the founders had already approved plans including the addition of chatbot features to Google. The company is planning to launch over 20 AI products and will showcase a version of Google with chatbot features.
All this coincides with Microsoft’s massive investment into AI and ChatGPT. Google has been working with the language model LaMDA (Language Model For Dialogue Applications), and with Anthropic in the ring, it remains to be seen how potent Google’s AI vision will be and whether it will dominate the AI industry.
