The world is changing at a dizzying pace and artificial intelligence (AI) is the best proof of this. Since the public launch in November of the ChatGPT text generator , the fastest app ever to reach 100 million users , the field is flourishing like never before, normalizing human-machine interaction and accelerating the marathon between giants like Microsoft or Google to colonize a market that promises billion-dollar benefits. There is no week in which new technologies are not presented whose impact can be transformative. Yet none had been as frantic and promising as this one.
The first earthquake was on Tuesday. OpenAI, the research laboratory responsible for ChatGPT, presented GPT-4, a new language model that will expand the data that this chatbot feeds on, giving it much more power and precision. Thus, you can submit the AI to a university exam and get a grade among the 10% of the best human students. In addition, this evolution will allow ChatGPT to not only perform tasks through the descriptions that the user writes, but it will also be able to work with image and sound . “It represents an important leap, it argues and simulates reasoning more brilliantly,” technological analyst Antonio Ortiz, author of the podcast on AI ‘Monos Estocásticos’, explained to El Periódico, from the Prensa Ibérica group.
Microsoft’s big gamble
As its name suggests, OpenAI used to reveal to the public how its innovations work, but that’s over now. “We were wrong, it doesn’t make sense to have open source ,” one of the founders of the startup, Ilya Sutskever, has confessed to ‘The Verge’. The presentation of GPT-4 was marked by opacity, a suspicion that is due to the fact that the company is no longer a non-profit organization, but a branch of Microsoft that seeks to do business. Thus, the improvements of this latest language model are only available to paying subscribers.
services. On Thursday, it extended the integration of this ‘co-pilot,’ as it’s been called, to all of its applications — Word , Excel , PowerPoint , Outlook, Teams, and the Power Platform—under the promise of “transforming the future of work.” That means the AI will be able to follow your orders and write a draft, send emails, calculate bills and make presentations in a matter of seconds. “Everything we do on the Internet will have a co-pilot,” says Ortiz, author of the Error 500 newsletter. “And if it speeds up productivity, it can have a brutal economic impact.”
If Microsoft’s valuation fell 30% last year so far in 2023, it has already risen by more than 15%, almost 10% just since Monday. The internal pressure strategy to bring these AI tools to users is paying off.
Google is not far behind
Although ChatGPT is in the hands of Microsoft, this chat system would not exist without the technology invented by Google. Until now, the Alphabet-owned giant had taken a more prudent course and kept its AI innovations under lock and key. However, the acceleration hit by its rival is seen as a possible threat to its business, which has forced the company to respond to keep up with the fast-paced competition.
At the beginning of January, Google launched Bard, its personal chatbot. Its use is, for the moment, reserved for a limited number of experts. On Tuesday he announced another battery of no less novelties. On the one hand, it will do like Microsoft and integrate generative AI into all its work products, such as Gmail, Docs, Drive or Meets. This technology —which is already being applied to auto-complete words or phrases— will make it possible to summarize conversations by email or take notes automatically during video calls.
On the other, it will open up access to its AI so that companies can use it to develop new services and applications. With this, it responds to OpenAI, whose business model happens to be the platform that enables new creations.
Also on Tuesday, AI company Anthropic — in which Google invested $400 million — released a new language model to compete with ChatGPT. Shares of the technology multinational have shot up 11.5% since Monday.
Companies from all over the world are signing up for this long-distance race. This Thursday, the Chinese company Baidu —known as the Chinese Google— presented Ernie Bot, a system with characteristics similar to GPT4 that will be limited to the Mandarin market. ” China has been very visionary for a decade , but its effort to control that these language models do not escape censorship may limit similar initiatives,” says Ortiz.
This week there have also been news beyond the text generators. The Midjourney research laboratory presented Wednesday Midjourney V5, its engine capable of generating all kinds of images from user descriptions. This evolution opens the door to much higher quality images . So much so that, in some cases, it becomes difficult to distinguish between real photos and what the machine has created. Ortiz points out that the next step will be in the video generators. “We are only seeing the sunrise,” he warns.