Google Bard was exposed to plagiarizing ChatGPT? BERT’s job-hopping OpenAI reveals the shocking inside story

Although Devlin switched to OpenAI in January this year, we knew it very early.

However, Bard’s training with ChatGPT data was only recently exposed.

According to The Information, one of the reasons why Devlin left Google was that he discovered that Bard, the seed player that Google used to fight against ChatGPT, was using ChatGPT data when training.

He warned CEO Pichai and other executives that the Bard team was training with information from ShareGPT and then resigned.

ShareGPT can generate a URL by capturing the complete conversation between others and ChatGPT. With this extension, people can directly share the URL, saving the time of taking screenshots of multiple conversations.

In other words, if Bard is trained with ShareGPT information, it is likely to train Bard into a “replica ChatGPT”.

According to insiders, after Devlin issued a warning, Google immediately stopped using this data to train Bard.

But when foreign media The Verge asked Google spokesman Chris Pappas about the incident, he denied it.

He said with certainty, “Bard’s training did not use any data from ShareGPT or ChatGPT.”

Is it useful or not? This matter is still a Rashomon.

Interestingly, OpenAI has often been involved in controversy before, with many websites and artists accusing ChatGPT of stealing data from itself.

And today, for the first time, someone has accused another company of stealing data from ChatGPT.

In other words, your circle is a circle.

Google’s counterattack is a bit embarrassing

Previously, the popular ChatGPT was integrated into Bing, which has already caused Google’s stock price to fall.

Google, whose status as a search giant has been shaken, is desperately trying to catch up.

As a result, Bard made a mistake on the spot at the press conference, making Google a laughing stock and evaporating 100 billion in market value.

After more than a month of secrecy, Bard finally officially opened. After using it, everyone found that the correct rate of Bard is not high, and it is not very good at writing codes. Compared with ChatGPT, it is more stretched.

The most critical problem is that Bard is already one step too late. Even if it was officially released, the media and the public have responded, and no one seems to care.

The choice of Devlin to leave Google and join OpenAI is even more meaningful-since Google relies so heavily on ShareGPT, the master simply joined OpenAI without a middleman.

What’s more serious is that Microsoft has an exclusive license to use ChatGPT for commercial purposes.

So Devlin is concerned that Google is using the data in this way, violating OpenAI’s terms of service. If the matter is confirmed, Google is likely to face lawsuits.

Of course, OpenAI is not innocent. The data for training ChatGPT is also “borrowed” from the public Internet. These contents are not licensed by human creators such as artists, writers, and videographers.

Though Microsoft did announce some compensation plans recently to compensate content creators who contribute to Bing Chat answers.

But at the end of the day, whether it’s Google or Microsoft, the only thing they really care about is profit for shareholders.

The Marriage of Google Brain and DeepMind

In addition to the big melon above, there is another melon in The Information’s report.

Since ChatGPT exploded in November last year, the arms race between Google and OpenAI has become increasingly tense.

Now, in order to fight against ChatGPT, Google’s two strongest “brains”-DeepMind and Google Brain, have chosen to join forces.

In recent weeks, DeepMind and Google Brain have been developing a rival to OpenAI, internally code-named Gemini, according to two people familiar with the matter.

Since Google acquired DeepMind in 2014, DeepMind has released several amazing artificial bits of intelligence, such as AlphaGO, AlphaCode, and AlphaFold. It can be said that the limelight far exceeds Google.

Since the release of ChatGPT, Google Brain has updated some progress from time to time, but DeepMind has been announcing recent big moves.

And now, employees at Alphabet’s two artificial intelligence labs believe that OpenAI has far surpassed them. In addition, many key engineers and researchers have flowed to OpenAI, making Google insiders worry.

In addition, after Bard was opened to some users last week, its performance was not as good as ChatGPT’s.

Seeing that the child Bard raised by himself was not satisfactory, Google researchers simply wanted another one, to be able to compete with GPT-4.

From this point of view, the importance of Gemini (Gemini) is self-evident.

One of the people familiar with the matter said that Jef Dean, the head of Google Brain and Google’s senior director of artificial intelligence research, has manually typed the code himself in the project.

It can be said that Gemini was born as a child of forced marriage. Because, previously, the two laboratories DeepMind and Google Brain rarely cooperated with each other or shared computer code.

But now, as both parties want to develop their own machine-learning models to compete with OpenAI and require a lot of computing power, a person familiar with the matter said. So they have no choice but to cooperate.

Today, the effort has seen Google’s countless teams in search and cloud, among other product lines, shaken by competition from ChatGPT.

No money to make, Google Assistant restructuring

No, just recently, the Google Assistant department is facing restructuring to focus on the development of Bard.

In a memo to employees, “Changes to Assistant and Bard teams,” a number of executive changes were listed.

A Google Assistant engineering vice president, Amar Subramanya, will now lead Bard. Another vice president of engineering, Mao Jianchang, left Google “due to personal reasons”, and Peeyush Ranjan, vice president of Google’s business department, became the taker.

“Given the tireless efforts of the Bard team, we look forward to continuing to support and execute future research projects,” Sissie Hsiao, vice president and head of the Google Assistant business unit, said in the memo.

It sounds like the Assistant team is now backing Google in the fight against ChatGPT “Code Red”.

Is Google Assistant doomed? In fact, there is growing evidence that the sector is on the decline:

• Google Assistant has released eight major speaker/smart display hardware in five years from 2016 to 2021, but hardware launches appear to have stalled. The last hardware launch was in March 2021. That was exactly two years ago.
• In 2022, Google removed Assistant support from two internal product lines: Nest Wi-Fi and Fitbit wearables.
• 2022 also saw a report from The Information that said Google wanted to “reduce investment in developing its Google Assistant voice-assisted search for use in cars and devices not made by Google.”
• Google Assistant’s driving mode turns off in 2022.
• Google Assistant’s “Duplex on the web” feature also shuts down in 2022.
• A core unique feature of the Google Assistant, Reminders, is shutting down in favor of Google Task Reminders.
• Google Assistant never made money. The hardware is sold at cost, there are no ads, and no one pays a monthly fee to use Google Assistant. Servers to process all those voice commands are also costly, although some newer devices have moved on-device processing in a veiled cost-cutting move. Assistant’s biggest competitor, Amazon Alexa, is also in the same boat, losing $10 billion a year.

It can also be seen from the development of Google Assistant over the years that the way Google operates is really too feudal.

The company seems to have a conventional view that making projects work on the back of existing products, such as maps or assistants, is a reliable way to make money.

So while the company has stockpiled many of the world’s best AI researchers, their talents seem to be trapped in the orbit of corporate strategy.

Before Bard, Google’s progress in the marketization of AI over the years was nothing more than continuous improvement of Google Assistant and other gadgets, and using AI to make some leftovers.

Such a giant should have followed the trend, promoted the trend, and stirred up the world.