We asked AI which striker Arsenal should sign in January – and the results were surprising

There are two great battles being played out in front our very eyes right now. On the one hand, the fight between man and machine, as tech bros work to code algorithms which could completely change the working landscape, for good and ill. On the other, the battle that Arsenal are in the midst of to sign a striker who knows where the net is. So what if we brought the two together?

Can AI really replace traditional analysts and journalists in the future? Will scouts be no more? That’s one potential future, but now we can pit our own work against the best the computers can do to find out.

Earlier this week, we provided our readers with a detailed breakdown of Arsenal’s likely options in the January transfer window, replete with statistical data and the latest news reports. But if we just ask a machine learning algorithm to do the same work, will they can up with better suggestions or not?

We asked AI to offer up some suggestions by bringing together data and news reports, just as we did. A real-life human being, with flesh, blood and mouths to feed, suggested that Viktor Gyökeres was the way to go, with Yoane Wissa a possible back-up. So let’s see how a computer that wants to take our man’s job answered some key questions, and figure out whether we should be dishing out any P45s.

Of the strikers linked with Arsenal, which would be the best signing?

The AI in question is pretty emphatic here – 21-year-old RB Leipzig striker Benjamin Šeško is the best man for the job. Apparently, his “unique blend of youth, potential, and suitability to Arsenal's current and future tactical setup”.

That last sentence scares us, to be honest. How does AI know Arsenal’s future tactics? Just what technology have these people got access to? How far down the road to our own ruin are we? Skynet had time travel, lest we forget.

Granted, it could just be the computerised equivalent of a slip of the tongue, so we’ll worry more about whether it’s right about Šeško for now. There’s no doubting Šeško’s potential, but with just eight goals in 17 league games this season, it’s debatable whether he’s ready to swing a title race just yet.

Still, as the AI observes, the Slovenian has “an ability to outperform his expected goals (xG), suggesting he's not just a scorer but also savvy in positioning and finishing”. That’s certainly true – those eight goals came from just 4.5xG, which is an obscene conversion rate, and he scored at much the same rate last season.

But should Arsenal be worried about a striker at a big club having such a low xG in the first place? 4.5 is pretty paltry, and could suggest that he doesn’t get into good enough positions yet. AI really doesn’t have any answers here – when pressed, it just offers an explanation of why a generic player might have a lower xG. We suspect Arteta might expect a more detailed answer from his scouting staff.

There is one problem, however – Šeško is almost certainly unavailable this winter, and they would have to wait until the summer, something the AI fails to make mention of but Fabrizio Romano does in his reporting. Still, the AI could be proven right in a few years’ time, so let’s not be too harsh.

Why would Šeško be better than Viktor Gyökeres?

When asked for a definitive, simple answer, the AI tells us flatly: “Benjamin Šeško would be a better signing for Arsenal.” But why is that? The answer, apparently, breaks down to two issues with the free-scoring Sporting forward. Apparently, Šeško’s “ability to play as a traditional striker with good hold-up play” and “capacity to operate in various tactical setups” makes him a better fit for Mikel Arteta’s system. Gyökeres, apparently, is “more fixed”, which we’re slightly dubious about.

The AI also seems fixated on xG conversion. Šeško “outperforms his xG, suggesting he can create and finish chances that aren't necessarily the highest quality”, which could be essential in tight games. Meanwhile, Gyökeres’ “scoring rate is very high… his xG might be more reflective of the chances he gets, indicating a different kind of impact.” We have no idea what that means.

Meanwhile, “Šeško’s background and the kind of player he's shown to be might align better with Arsenal's ethos”. Fair enough, perhaps, but is it fair to say that Gyökeres’ “integration into Arsenal's squad culture might take more time, given he's coming from a different football culture”? He’s actually played in England for a few years after all.

All in all, we’re sticking with our judgement that Gyökeres scores more, is a better all-round player right now, is no less likely to fit in and still has a really, really good xG conversion rate, but come back in a few years, perhaps.

If Arsenal can’t sign Gyökeres or Šeško, who else should they consider?

To get cleaner answers, we gave the AI prompts to look at cheaper players, assuming that the Gunners would want to go back in for one of their top targets later on while adding someone this winter to get the goals required to compete in the title race. The answers it provided were… of mixed quality, in our opinion, but see what you think.

It kicks off by mentioning Feyenoord’s Santiago Gimnénez, which is a solid start – he has a big wobble form-wise around a year ago but has recovered and posted seven goals in his first 10 Eredivise games so far. He’s big, physical, and scores goals. A solid option.

Jhon Durán gets a nod, which doesn’t really seem to fit the “cheap” bill hardly looks like a terrible plan based on this season’s form in front of goal, as does Ademola Lookman, although we question whether you can define him as a striker. Surely Arsenal need someone a bit more central.

So far, the AI isn’t doing too badly, but the suggestion that really made us miss the early days of AI, when it would apologise politely when it came up with something really stupid, was Armando Broja.

Yes, the Chelsea forward who is allegedly on loan at Everton but hasn’t been seen for some time due to a string of serious injuries. Apparently, he would be a “low-cost, high-reward option” while his “physical style and ability in the air” could work for Arteta. Maybe, but let’s be serious for a minute – Broja isn’t even fit right now, less still in any fit state to lead a title challenge.

So what do we make of the results? Well, it looks like the future employability of human journalists may hinge on whether Gyökeres or Šeško are the better striker in a few years’ time. We suspect that the machines can’t quite grasp the concept of short-termism at the moment, and even when prompted to consider this year’s title race, it still backed the Slovenian with some increasingly dubious logic. Maybe we’re wrong. We’re only human, after all. But given that it came up with Broja instead of Wissa, we’re going to suggest you still don’t use it to do your homework.