ChatGPT ‘Outmatched’ by a Nearly 50-Year-Old Console!

Be careful, old -fashioned. In 1977, the Atari 2600 console appeared on the stalls of video game merchants. It was almost 50 years ago! Two years later, the game Atari Chess officially came out. At that time, to support again where it hurts, the President of the Republic was Valéry Giscard d'Estaing. Well Chatgpt, the most advanced language model in the world, capable of brilliantly passing the Turing test, has lamentedly lost to failures against this console.

explained on LinkedIn how the language model was “demolish at the beginner level”. High stroke.

The basic experience wanted to see the speed at which Chatgpt was going to beat a game system which only had ahead of chess. Well, it seems that Chatgpt muscle his game because in addition to having lost, the AI ​​even tried to find excuses by explaining that the icons of the playing platforms were “too abstract to be recognizable”. The AI ​​would therefore have also developed the concept of bad faith. Fascinating.

To allow Chatgpt to have all the cards in hand to succeed, the engineer would even have put the game in standard failure notation. Result: Chatgpt has lost again. In short, the problem obviously did not come from the parts and the Atari 2600 was visibly too strong for the AI.

AI and chess

When these two words are mentioned in the same sentence, a name necessarily comes to mind: Garry Kasparov. Undisputed legend of failures, long before the advent of Magnus Carlsen, he distinguished himself for having faced many computers who were designed to beat him.

But two games are still in all the heads of those who are a fan of chess or AI, like Chatgpt,: Kasparov vs Deep Blue (1996) and Kasparov vs Deeper Blue (1997). The experience was simple, the superordinators developed by IBM had to try to beat Garry Kasparov, still considered today as one of the best chess players of all time.

Thus, between February 10 and 17, 1996, the first match between Deep Blue and Kasparov is played out of which here is the course:

  • Part 1 – Deep Blue
  • Part 2 – Kasparov
  • Party 3 – Null
  • Part 4 – Null
  • Part 5 – Kasparov
  • Part 6 – Kasparov

Result of the match: 4-2 for Kasparov.

A year later, from May 3 to 11, 1997, revenge is then played against a more optimized computer and more able to beat the legend: Deeper Blue. The match takes place as follows:

  • Part 1 – Kasparov
  • Part 2 – Deepper Blue
  • Part 3 – Null
  • Part 4 – Null
  • Part 5 – Null
  • Part 6 – Deepper Blue

Result: 3.5-2.5 to Deep Blue.

And that will not please multiple Russian world champion and the first player in history to have exceeded the Elo 2800. Indeed, he challenged the victory of the machine by explaining that the computer had had access to the parts he had played before to train, something that could not have had Kasparov.

Anyway, the bet for IBM was successful, his supermarket had managed to beat the legend. Today, who has become pariah in his country and even considered a “terrorist” in the eyes of the Kremlin since 2024, he has no right to return to Russia and fills fate, he works today in the development of artificial intelligence.

Perhaps he will be able to give a little help to Chatgpt to help him, finally, beat the Atari 2600 beyond the beginner level.

Source : Futurism

More news

Berlin’s Unsold Christmas Trees Repurposed to Nourish Zoo Elephants

Even after the holidays, the Christmas spirit continues to be felt at Berlin Zoo. To the delight of the park animals, it was time ...

Concerned About Authoritarian Trends, Researchers Are Leaving OpenAI in Droves

When technologies advance at full speed, transparency becomes just as essential as innovation. In the field of artificial intelligence, it is sometimes the researchers ...

Resurrected from the Depths: The French Submarine Le Tonnant, Lost in 1942, Unearths a Forgotten Chapter of WWII off Spain’s Coast

For more than eight decades, Le Tonnant existed only in military reports and family memories. Scuttled in the chaos of the Second World War, ...

Leave a Comment